Improve accessibility of microphone, camera, and screen share buttons

Taking Valere's suggestion of giving them the 'switch' role. Also, the aria-label attributes were redundant (having tooltips already gives the buttons aria-labelledby).

.github/workflows/blocked.yaml

@@ -1,7 +1,16 @@
 name: Prevent blocked
 on:
+  # zizmor: ignore[dangerous-triggers]
+  # Reason: This workflow does not checkout code or use secrets.
+  # It only reads labels to set a failure status on the PR.
   pull_request_target:
     types: [opened, labeled, unlabeled, synchronize]
+
+permissions:
+  pull-requests: read
+  # Required to fail the check on the PR
+  statuses: write
+
 jobs:
   prevent-blocked:
     name: Prevent blocked

.github/workflows/build-and-publish-docker.yaml

@@ -25,6 +25,8 @@ jobs:
     steps:
       - name: Check it out
         uses: actions/checkout@34e114876b0b11c390a56381ad16ebd13914f8d5 # v4
+        with:
+          persist-credentials: false
 
       - name: 📥 Download artifact
         uses: actions/download-artifact@d3f86a106a0bac45b974a628896c90dbdf5c8093 # v4
@@ -42,7 +44,7 @@ jobs:
           password: ${{ secrets.GITHUB_TOKEN }}
 
       - name: Connect to Tailscale
-        uses: tailscale/github-action@53acf823325fe9ca47f4cdaa951f90b4b0de5bb9 # v4
+        uses: tailscale/github-action@306e68a486fd2350f2bfc3b19fcd143891a4a2d8 # v4
         if: github.event_name != 'pull_request'
         with:
           oauth-client-id: ${{ secrets.TS_OAUTH_CLIENT_ID }}
@@ -66,8 +68,8 @@ jobs:
           jwtGithubAudience: https://vault.infra.ci.i.element.dev
           method: jwt
           secrets: |
-            services/<team>-repositories/secret/data/oci.element.io username | OCI_USERNAME ;
-            services/<team>-repositories/secret/data/oci.element.io password | OCI_PASSWORD ;
+            services/voip-repositories/secret/data/oci.element.io username | OCI_USERNAME ;
+            services/voip-repositories/secret/data/oci.element.io password | OCI_PASSWORD ;
 
       - name: Login to oci.element.io Registry
         uses: docker/login-action@c94ce9fb468520275223c153574b00df6fe4bcc9 # v3
@@ -83,7 +85,7 @@ jobs:
         with:
           images: |
             ${{ env.REGISTRY }}/${{ env.IMAGE_NAME }}
-            oci-push.vpn.infra.element.io/element-web
+            oci-push.vpn.infra.element.io/element-call
           tags: ${{ inputs.docker_tags }}
           labels: |
             org.opencontainers.image.licenses=AGPL-3.0-only OR LicenseRef-Element-Commercial

.github/workflows/build-element-call.yaml

@@ -33,6 +33,8 @@ jobs:
     steps:
       - name: Checkout code
         uses: actions/checkout@34e114876b0b11c390a56381ad16ebd13914f8d5 # v4
+        with:
+          persist-credentials: false
       - name: Enable Corepack
         run: corepack enable
       - name: Yarn cache
@@ -43,7 +45,7 @@ jobs:
       - name: Install dependencies
         run: "yarn install --immutable"
       - name: Build Element Call
-        run: ${{ format('yarn run build:{0}:{1}', inputs.package, inputs.build_mode) }}
+        run: yarn run build:"$PACKAGE":"$BUILD_MODE"
         env:
           SENTRY_ORG: ${{ secrets.SENTRY_ORG }}
           SENTRY_PROJECT: ${{ secrets.SENTRY_PROJECT }}
@@ -52,6 +54,8 @@ jobs:
           CODECOV_TOKEN: ${{ secrets.CODECOV_TOKEN }}
           VITE_APP_VERSION: ${{ inputs.vite_app_version }}
           NODE_OPTIONS: "--max-old-space-size=4096"
+          PACKAGE: ${{ inputs.package }}
+          BUILD_MODE: ${{ inputs.build_mode }}
       - name: Upload Artifact
         uses: actions/upload-artifact@ea165f8d65b6e75b540449e92b4886f43607fa02 # v4
         with:

.github/workflows/build.yaml

@@ -51,6 +51,7 @@ jobs:
       packages: write
       id-token: write
     uses: ./.github/workflows/build-and-publish-docker.yaml
+    secrets: inherit
     with:
       artifact_run_id: ${{ github.run_id }}
       docker_tags: |

.github/workflows/changelog-label.yml

@@ -1,8 +1,16 @@
 name: PR changelog label
 
 on:
+  # zizmor: ignore[dangerous-triggers]
+  # This is safe because we do not use actions/checkout or execute untrusted code.
+  # Using pull_request_target is necessary to allow status writes for PRs from forks.
   pull_request_target:
     types: [labeled, unlabeled, opened]
+
+permissions:
+  pull-requests: read
+  statuses: write
+
 jobs:
   pr-changelog-label:
     runs-on: ubuntu-latest

.github/workflows/deploy-to-netlify.yaml

@@ -63,11 +63,15 @@ jobs:
 
       - name: Add config file
         run: |
-          if [ "${{ inputs.package }}" = "full" ]; then
-            curl -s "https://raw.githubusercontent.com/${{ inputs.pr_head_full_name }}/${{ inputs.pr_head_ref }}/config/config_netlify_preview.json" > webapp/config.json
+          if [ "${INPUTS_PACKAGE}" = "full" ]; then
+            curl -s "https://raw.githubusercontent.com/${INPUTS_PR_HEAD_FULL_NAME}/${INPUTS_PR_HEAD_REF}/config/config_netlify_preview.json" > webapp/config.json
           else
-            curl -s "https://raw.githubusercontent.com/${{ inputs.pr_head_full_name }}/${{ inputs.pr_head_ref }}/config/config_netlify_preview_sdk.json" > webapp/config.json
+            curl -s "https://raw.githubusercontent.com/${INPUTS_PR_HEAD_FULL_NAME}/${INPUTS_PR_HEAD_REF}/config/config_netlify_preview_sdk.json" > webapp/config.json
           fi
+        env:
+          INPUTS_PACKAGE: ${{ inputs.package }}
+          INPUTS_PR_HEAD_FULL_NAME: ${{ inputs.pr_head_full_name }}
+          INPUTS_PR_HEAD_REF: ${{ inputs.pr_head_ref }}
       - name: ☁️ Deploy to Netlify
         id: netlify
         uses: nwtgck/actions-netlify@4cbaf4c08f1a7bfa537d6113472ef4424e4eb654 # v3.0

.github/workflows/lint.yaml

@@ -8,6 +8,8 @@ jobs:
     steps:
       - name: Checkout code
         uses: actions/checkout@34e114876b0b11c390a56381ad16ebd13914f8d5 # v4
+        with:
+          persist-credentials: false
       - name: Enable Corepack
         run: corepack enable
       - name: Yarn cache

.github/workflows/pr-deploy.yaml

@@ -1,5 +1,7 @@
 name: Deploy previews for PRs
 on:
+  # zizmor: ignore[dangerous-triggers]
+  # Reason: This is now restricted to internal PRs only using the 'if' condition below.
   workflow_run:
     workflows: ["Build"]
     types:
@@ -7,7 +9,14 @@ on:
 
 jobs:
   prdetails:
-    if: ${{ github.event.workflow_run.conclusion == 'success' && github.event.workflow_run.event == 'pull_request' }}
+    # Logic:
+    # 1. Build must be successful
+    # 2. Event must be a pull_request
+    # 3. Head repository must be the SAME as the base repository (No Forks!)
+    if: >
+      github.event.workflow_run.conclusion == 'success' && 
+      github.event.workflow_run.event == 'pull_request' &&
+      github.event.workflow_run.head_repository.full_name == github.repository
     runs-on: ubuntu-latest
     outputs:
       pr_number: ${{ steps.prdetails.outputs.pr_id }}
@@ -62,6 +71,7 @@ jobs:
       packages: write
       id-token: write
     uses: ./.github/workflows/build-and-publish-docker.yaml
+    secrets: inherit
     with:
       artifact_run_id: ${{ github.event.workflow_run.id || github.run_id }}
       docker_tags: |

.github/workflows/publish-embedded-packages.yaml

@@ -22,8 +22,18 @@ jobs:
       TAG: ${{ steps.tag.outputs.TAG }}
     steps:
       - name: Calculate VERSION
-        # We should only use the hard coded test value for a dry run
-        run: echo "VERSION=${{ github.event_name == 'release' && github.event.release.tag_name || 'v0.0.0-pre.0' }}" >> "$GITHUB_ENV"
+        # Safely store dynamic values in environment variables
+        # to prevent shell injection (template-injection)
+        run: |
+          # The logic is executed within the shell using the env variables
+          if [ "$EVENT_NAME" = "release" ]; then
+            echo "VERSION=$RELEASE_TAG" >> "$GITHUB_ENV"
+          else
+            echo "VERSION=v0.0.0-pre.0" >> "$GITHUB_ENV"
+          fi
+        env:
+          RELEASE_TAG: ${{ github.event.release.tag_name }}
+          EVENT_NAME: ${{ github.event_name }}
       - id: dry_run
         name: Set DRY_RUN
         # We perform a dry run for all events except releases.
@@ -71,7 +81,9 @@ jobs:
       contents: write # required to upload release asset
     steps:
       - name: Determine filename
-        run: echo "FILENAME_PREFIX=element-call-embedded-${{ needs.versioning.outputs.UNPREFIXED_VERSION }}" >> "$GITHUB_ENV"
+        run: echo "FILENAME_PREFIX=element-call-embedded-${NEEDS_VERSIONING_OUTPUTS_UNPREFIXED_VERSION}" >> "$GITHUB_ENV"
+        env:
+          NEEDS_VERSIONING_OUTPUTS_UNPREFIXED_VERSION: ${{ needs.versioning.outputs.UNPREFIXED_VERSION }}
       - name: 📥 Download built element-call artifact
         uses: actions/download-artifact@d3f86a106a0bac45b974a628896c90dbdf5c8093 # v4
         with:
@@ -80,9 +92,9 @@ jobs:
           name: build-output-embedded
           path: ${{ env.FILENAME_PREFIX}}
       - name: Create Tarball
-        run: tar --numeric-owner -cvzf ${{ env.FILENAME_PREFIX }}.tar.gz ${{ env.FILENAME_PREFIX }}
+        run: tar --numeric-owner -cvzf ${FILENAME_PREFIX}.tar.gz ${FILENAME_PREFIX}
       - name: Create Checksum
-        run: find ${{ env.FILENAME_PREFIX }} -type f -print0 | sort -z | xargs -0 sha256sum | tee ${{ env.FILENAME_PREFIX }}.sha256
+        run: find ${FILENAME_PREFIX} -type f -print0 | sort -z | xargs -0 sha256sum | tee ${FILENAME_PREFIX}.sha256
       - name: Upload
         if: ${{ needs.versioning.outputs.DRY_RUN == 'false' }}
         uses: softprops/action-gh-release@a06a81a03ee405af7f2048a818ed3f03bbf83c7b # v2
@@ -104,6 +116,8 @@ jobs:
     steps:
       - name: Checkout
         uses: actions/checkout@34e114876b0b11c390a56381ad16ebd13914f8d5 # v4
+        with:
+          persist-credentials: false
 
       - name: 📥 Download built element-call artifact
         uses: actions/download-artifact@d3f86a106a0bac45b974a628896c90dbdf5c8093 # v4
@@ -123,13 +137,16 @@ jobs:
       - name: Publish npm
         working-directory: embedded/web
         run: |
-          npm version ${{ needs.versioning.outputs.PREFIXED_VERSION }} --no-git-tag-version
+          npm version ${NEEDS_VERSIONING_OUTPUTS_PREFIXED_VERSION} --no-git-tag-version
           echo "ARTIFACT_VERSION=$(jq '.version' --raw-output package.json)" >> "$GITHUB_ENV"
-          npm publish --provenance --access public --tag ${{ needs.versioning.outputs.TAG }} ${{ needs.versioning.outputs.DRY_RUN == 'true' && '--dry-run' || '' }}
+          npm publish --provenance --access public --tag ${NEEDS_VERSIONING_OUTPUTS_TAG} ${{ needs.versioning.outputs.DRY_RUN == 'true' && '--dry-run' || '' }}
+        env:
+          NEEDS_VERSIONING_OUTPUTS_PREFIXED_VERSION: ${{ needs.versioning.outputs.PREFIXED_VERSION }}
+          NEEDS_VERSIONING_OUTPUTS_TAG: ${{ needs.versioning.outputs.TAG }}
 
       - id: artifact_version
         name: Output artifact version
-        run: echo "ARTIFACT_VERSION=${{env.ARTIFACT_VERSION}}" >> "$GITHUB_OUTPUT"
+        run: echo "ARTIFACT_VERSION=${ARTIFACT_VERSION}" >> "$GITHUB_OUTPUT"
 
   publish_android:
     needs: [build_element_call, versioning]
@@ -143,6 +160,8 @@ jobs:
     steps:
       - name: Checkout
         uses: actions/checkout@34e114876b0b11c390a56381ad16ebd13914f8d5 # v4
+        with:
+          persist-credentials: false
 
       - name: 📥 Download built element-call artifact
         uses: actions/download-artifact@d3f86a106a0bac45b974a628896c90dbdf5c8093 # v4
@@ -161,16 +180,19 @@ jobs:
       - name: Get artifact version
         # Anything that is not a final release will be tagged as a snapshot
         run: |
-          if [[ "${{ needs.versioning.outputs.TAG }}" == "latest" ]]; then
-            echo "ARTIFACT_VERSION=${{ needs.versioning.outputs.UNPREFIXED_VERSION }}" >> "$GITHUB_ENV"
-          elif [[ "${{ needs.versioning.outputs.TAG }}" == "rc" ]]; then
-            echo "ARTIFACT_VERSION=${{ needs.versioning.outputs.UNPREFIXED_VERSION }}" >> "$GITHUB_ENV"
+          if [[ "${NEEDS_VERSIONING_OUTPUTS_TAG}" == "latest" ]]; then
+            echo "ARTIFACT_VERSION=${NEEDS_VERSIONING_OUTPUTS_UNPREFIXED_VERSION}" >> "$GITHUB_ENV"
+          elif [[ "${NEEDS_VERSIONING_OUTPUTS_TAG}" == "rc" ]]; then
+            echo "ARTIFACT_VERSION=${NEEDS_VERSIONING_OUTPUTS_UNPREFIXED_VERSION}" >> "$GITHUB_ENV"
           else
-            echo "ARTIFACT_VERSION=${{ needs.versioning.outputs.UNPREFIXED_VERSION }}-SNAPSHOT" >> "$GITHUB_ENV"
+            echo "ARTIFACT_VERSION=${NEEDS_VERSIONING_OUTPUTS_UNPREFIXED_VERSION}-SNAPSHOT" >> "$GITHUB_ENV"
           fi
+        env:
+          NEEDS_VERSIONING_OUTPUTS_TAG: ${{ needs.versioning.outputs.TAG }}
+          NEEDS_VERSIONING_OUTPUTS_UNPREFIXED_VERSION: ${{ needs.versioning.outputs.UNPREFIXED_VERSION }}
 
       - name: Set version string
-        run: sed -i "s/0.0.0/${{ env.ARTIFACT_VERSION }}/g" embedded/android/lib/src/main/kotlin/io/element/android/call/embedded/Version.kt
+        run: sed -i "s/0.0.0/${ARTIFACT_VERSION}/g" embedded/android/lib/src/main/kotlin/io/element/android/call/embedded/Version.kt
 
       - name: Publish AAR
         working-directory: embedded/android
@@ -184,7 +206,7 @@ jobs:
 
       - id: artifact_version
         name: Output artifact version
-        run: echo "ARTIFACT_VERSION=${{env.ARTIFACT_VERSION}}" >> "$GITHUB_OUTPUT"
+        run: echo "ARTIFACT_VERSION=${ARTIFACT_VERSION}" >> "$GITHUB_OUTPUT"
 
   publish_ios:
     needs: [build_element_call, versioning]
@@ -200,6 +222,7 @@ jobs:
         uses: actions/checkout@34e114876b0b11c390a56381ad16ebd13914f8d5 # v4
         with:
           path: element-call
+          persist-credentials: false
 
       - name: 📥 Download built element-call artifact
         uses: actions/download-artifact@d3f86a106a0bac45b974a628896c90dbdf5c8093 # v4
@@ -215,15 +238,18 @@ jobs:
           repository: element-hq/element-call-swift
           path: element-call-swift
           token: ${{ secrets.SWIFT_RELEASE_TOKEN }}
+          persist-credentials: false
 
       - name: Copy files
         run: rsync -a --delete --exclude .git element-call/embedded/ios/ element-call-swift
 
       - name: Get artifact version
-        run: echo "ARTIFACT_VERSION=${{ needs.versioning.outputs.UNPREFIXED_VERSION }}" >> "$GITHUB_ENV"
+        run: echo "ARTIFACT_VERSION=${NEEDS_VERSIONING_OUTPUTS_UNPREFIXED_VERSION}" >> "$GITHUB_ENV"
+        env:
+          NEEDS_VERSIONING_OUTPUTS_UNPREFIXED_VERSION: ${{ needs.versioning.outputs.UNPREFIXED_VERSION }}
 
       - name: Set version string
-        run: sed -i "s/0.0.0/${{ env.ARTIFACT_VERSION }}/g" element-call-swift/Sources/EmbeddedElementCall/EmbeddedElementCall.swift
+        run: sed -i "s/0.0.0/${ARTIFACT_VERSION}/g" element-call-swift/Sources/EmbeddedElementCall/EmbeddedElementCall.swift
 
       - name: Test build
         working-directory: element-call-swift
@@ -235,17 +261,22 @@ jobs:
           git config --global user.email "ci@element.io"
           git config --global user.name "Element CI"
           git add -A
-          git commit -am "Release ${{ needs.versioning.outputs.PREFIXED_VERSION }}"
-          git tag -a ${{ env.ARTIFACT_VERSION }} -m "${{ github.event.release.html_url }}"
+          git commit -am "Release ${NEEDS_VERSIONING_OUTPUTS_PREFIXED_VERSION}"
+          git tag -a ${ARTIFACT_VERSION} -m "${GITHUB_EVENT_RELEASE_HTML_URL}"
+        env:
+          NEEDS_VERSIONING_OUTPUTS_PREFIXED_VERSION: ${{ needs.versioning.outputs.PREFIXED_VERSION }}
+          GITHUB_EVENT_RELEASE_HTML_URL: ${{ github.event.release.html_url }}
 
       - name: Push
         working-directory: element-call-swift
         run: |
-          git push --tags ${{ needs.versioning.outputs.DRY_RUN == 'true' && '--dry-run' || '' }}
+          git push "https://x-access-token:${SWIFT_RELEASE_TOKEN}@github.com/element-hq/element-call-swift.git" --tags ${{ needs.versioning.outputs.DRY_RUN == 'true' && '--dry-run' || '' }}
+        env:
+          SWIFT_RELEASE_TOKEN: ${{ secrets.SWIFT_RELEASE_TOKEN }}
 
       - id: artifact_version
         name: Output artifact version
-        run: echo "ARTIFACT_VERSION=${{env.ARTIFACT_VERSION}}" >> "$GITHUB_OUTPUT"
+        run: echo "ARTIFACT_VERSION=${ARTIFACT_VERSION}" >> "$GITHUB_OUTPUT"
 
   release_notes:
     needs: [versioning, publish_npm, publish_android, publish_ios]
@@ -257,9 +288,13 @@ jobs:
     steps:
       - name: Log versions
         run: |
-          echo "NPM: ${{ needs.publish_npm.outputs.ARTIFACT_VERSION }}"
-          echo "Android: ${{ needs.publish_android.outputs.ARTIFACT_VERSION }}"
-          echo "iOS: ${{ needs.publish_ios.outputs.ARTIFACT_VERSION }}"
+          echo "NPM: ${NEEDS_PUBLISH_NPM_OUTPUTS_ARTIFACT_VERSION}"
+          echo "Android: ${NEEDS_PUBLISH_ANDROID_OUTPUTS_ARTIFACT_VERSION}"
+          echo "iOS: ${NEEDS_PUBLISH_IOS_OUTPUTS_ARTIFACT_VERSION}"
+        env:
+          NEEDS_PUBLISH_NPM_OUTPUTS_ARTIFACT_VERSION: ${{ needs.publish_npm.outputs.ARTIFACT_VERSION }}
+          NEEDS_PUBLISH_ANDROID_OUTPUTS_ARTIFACT_VERSION: ${{ needs.publish_android.outputs.ARTIFACT_VERSION }}
+          NEEDS_PUBLISH_IOS_OUTPUTS_ARTIFACT_VERSION: ${{ needs.publish_ios.outputs.ARTIFACT_VERSION }}
       - name: Add release notes
         if: ${{ needs.versioning.outputs.DRY_RUN == 'false' }}
         uses: softprops/action-gh-release@a06a81a03ee405af7f2048a818ed3f03bbf83c7b # v2

.github/workflows/publish.yaml

@@ -38,9 +38,9 @@ jobs:
           name: build-output-full
           path: ${{ env.FILENAME_PREFIX }}
       - name: Create Tarball
-        run: tar --numeric-owner --transform "s/dist/${{ env.FILENAME_PREFIX }}/" -cvzf ${{ env.FILENAME_PREFIX }}.tar.gz ${{ env.FILENAME_PREFIX }}
+        run: tar --numeric-owner --transform "s/dist/${FILENAME_PREFIX}/" -cvzf ${FILENAME_PREFIX}.tar.gz ${FILENAME_PREFIX}
       - name: Create Checksum
-        run: find ${{ env.FILENAME_PREFIX }} -type f -print0 | sort -z | xargs -0 sha256sum | tee ${{ env.FILENAME_PREFIX }}.sha256
+        run: find ${FILENAME_PREFIX} -type f -print0 | sort -z | xargs -0 sha256sum | tee ${FILENAME_PREFIX}.sha256
       - name: Upload
         uses: softprops/action-gh-release@a06a81a03ee405af7f2048a818ed3f03bbf83c7b # v2
         with:
@@ -57,6 +57,7 @@ jobs:
       packages: write
       id-token: write
     uses: ./.github/workflows/build-and-publish-docker.yaml
+    secrets: inherit
     with:
       artifact_run_id: ${{ github.event.workflow_run.id || github.run_id }}
       docker_tags: |

.github/workflows/test.yaml

@@ -10,6 +10,8 @@ jobs:
     steps:
       - name: Checkout code
         uses: actions/checkout@34e114876b0b11c390a56381ad16ebd13914f8d5 # v4
+        with:
+          persist-credentials: false
       - name: Enable Corepack
         run: corepack enable
       - name: Yarn cache
@@ -34,6 +36,8 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - uses: actions/checkout@34e114876b0b11c390a56381ad16ebd13914f8d5 # v4
+        with:
+          persist-credentials: false
       - name: Enable Corepack
         run: corepack enable
       - uses: actions/setup-node@49933ea5288caeca8642d1e84afbd3f7d6820020 # v4

.github/workflows/translations-download.yaml

@@ -14,6 +14,8 @@ jobs:
     steps:
       - name: Checkout the code
         uses: actions/checkout@34e114876b0b11c390a56381ad16ebd13914f8d5 # v4
+        with:
+          persist-credentials: false
 
       - name: Enable Corepack
         run: corepack enable

.github/workflows/translations-upload.yaml

@@ -15,6 +15,8 @@ jobs:
     steps:
       - name: Checkout the code
         uses: actions/checkout@34e114876b0b11c390a56381ad16ebd13914f8d5 # v4
+        with:
+          persist-credentials: false
 
       - name: Upload
         uses: localazy/upload@27e6b5c0fddf4551596b42226b1c24124335d24a # v1

.github/workflows/zizmor.yml

@@ -0,0 +1,23 @@
+name: GitHub Actions Security Analysis with zizmor 🌈
+
+on:
+  push:
+    branches: ["livekit", "full-mesh"]
+  pull_request: {}
+
+permissions: {}
+
+jobs:
+  zizmor:
+    name: Run zizmor 🌈
+    runs-on: ubuntu-latest
+    permissions:
+      security-events: write # Required for upload-sarif (used by zizmor-action) to upload SARIF files.
+    steps:
+      - name: Checkout repository
+        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
+        with:
+          persist-credentials: false
+
+      - name: Run zizmor 🌈
+        uses: zizmorcore/zizmor-action@71321a20a9ded102f6e9ce5718a2fcec2c4f70d8 # v0.5.2

knip.ts

@@ -34,6 +34,12 @@ export default {
     // then Knip will flag it as a false positive
     // https://github.com/webpro-nl/knip/issues/766
     "@vector-im/compound-web",
+    // Yarn plugins are allowed to depend on packages provided by the Yarn
+    // runtime. These shouldn't be listed in package.json, because plugins
+    // should work before Yarn even installs dependencies for the first time.
+    // https://yarnpkg.com/advanced/plugin-tutorial#what-does-a-plugin-look-like
+    "@yarnpkg/core",
+    "@yarnpkg/parsers",
     "matrix-widget-api",
   ],
   ignoreExportsUsedInFile: true,

locales/en/app.json

@@ -250,11 +250,11 @@
   "video_tile": {
     "always_show": "Always show",
     "camera_starting": "Video loading...",
-    "change_fit_contain": "Fit to frame",
     "collapse": "Collapse",
     "expand": "Expand",
     "mute_for_me": "Mute for me",
     "muted_for_me": "Muted for me",
+    "screen_share_volume": "Screen share volume",
     "volume": "Volume",
     "waiting_for_media": "Waiting for media..."
   }

package.json

@@ -48,7 +48,7 @@
     "@livekit/components-core": "^0.12.0",
     "@livekit/components-react": "^2.0.0",
     "@livekit/protocol": "^1.42.2",
-    "@livekit/track-processors": "^0.6.0 || ^0.7.1",
+    "@livekit/track-processors": "^0.7.1",
     "@mediapipe/tasks-vision": "^0.10.18",
     "@playwright/test": "^1.57.0",
     "@radix-ui/react-dialog": "^1.0.4",
@@ -79,7 +79,7 @@
     "@vector-im/compound-design-tokens": "^6.0.0",
     "@vector-im/compound-web": "^8.0.0",
     "@vitejs/plugin-react": "^4.0.1",
-    "@vitest/coverage-v8": "^3.0.0",
+    "@vitest/coverage-v8": "^4.0.18",
     "babel-plugin-transform-vite-meta-env": "^1.0.3",
     "classnames": "^2.3.1",
     "copy-to-clipboard": "^3.3.3",
@@ -101,7 +101,7 @@
     "i18next-browser-languagedetector": "^8.0.0",
     "i18next-parser": "^9.1.0",
     "jsdom": "^26.0.0",
-    "knip": "5.82.1",
+    "knip": "^5.86.0",
     "livekit-client": "^2.13.0",
     "lodash-es": "^4.17.21",
     "loglevel": "^1.9.1",
@@ -128,17 +128,22 @@
     "unique-names-generator": "^4.6.0",
     "uuid": "^13.0.0",
     "vaul": "^1.0.0",
-    "vite": "^7.0.0",
+    "vite": "^7.3.0",
     "vite-plugin-generate-file": "^0.3.0",
     "vite-plugin-html": "^3.2.2",
     "vite-plugin-node-stdlib-browser": "^0.2.1",
     "vite-plugin-svgr": "^4.0.0",
-    "vitest": "^3.0.0",
+    "vitest": "^4.0.18",
     "vitest-axe": "^1.0.0-pre.3"
   },
   "resolutions": {
     "@livekit/components-core/rxjs": "^7.8.1",
-    "@livekit/track-processors/@mediapipe/tasks-vision": "^0.10.18"
+    "@livekit/track-processors/@mediapipe/tasks-vision": "^0.10.18",
+    "minimatch": "^10.2.3",
+    "tar": "^7.5.11",
+    "glob": "^10.5.0",
+    "qs": "^6.14.1",
+    "js-yaml": "^4.1.1"
   },
   "packageManager": "yarn@4.7.0"
 }

playwright/create-call.spec.ts

@@ -22,8 +22,8 @@ test("Start a new call then leave and show the feedback screen", async ({
   await expect(page.getByTestId("lobby_joinCall")).toBeVisible();
 
   // Check the button toolbar
-  // await expect(page.getByRole('button', { name: 'Mute microphone' })).toBeVisible();
-  // await expect(page.getByRole('button', { name: 'Stop video' })).toBeVisible();
+  // await expect(page.getByRole('switch', { name: 'Mute microphone' })).toBeVisible();
+  // await expect(page.getByRole('switch', { name: 'Stop video' })).toBeVisible();
   await expect(page.getByRole("button", { name: "Settings" })).toBeVisible();
   await expect(page.getByRole("button", { name: "End call" })).toBeVisible();
 

playwright/reconnect.spec.ts

@@ -49,12 +49,12 @@ test("can only interact with header and footer while reconnecting", async ({
   ).toBeVisible();
 
   // Tab order should jump directly from header to footer, skipping media tiles
-  await page.getByRole("button", { name: "Mute microphone" }).focus();
+  await page.getByRole("switch", { name: "Mute microphone" }).focus();
   await expect(
-    page.getByRole("button", { name: "Mute microphone" }),
+    page.getByRole("switch", { name: "Mute microphone" }),
   ).toBeFocused();
   await page.keyboard.press("Tab");
-  await expect(page.getByRole("button", { name: "Stop video" })).toBeFocused();
+  await expect(page.getByRole("switch", { name: "Stop video" })).toBeFocused();
   // Most critically, we should be able to press the hangup button
   await page.getByRole("button", { name: "End call" }).click();
 });

playwright/widget/huddle-call.test.ts

@@ -55,13 +55,10 @@ widgetTest("Create and join a group call", async ({ addUser, browserName }) => {
     const frame = user.page
       .locator('iframe[title="Element Call"]')
       .contentFrame();
-
     // No lobby, should start with video on
-    // The only way to know if it is muted or not is to look at the data-kind attribute..
-    const videoButton = frame.getByTestId("incall_videomute");
-    await expect(videoButton).toBeVisible();
-    // video should be off by default in a voice call
-    await expect(videoButton).toHaveAttribute("aria-label", /^Stop video$/);
+    await expect(
+      frame.getByRole("switch", { name: "Stop video", checked: true }),
+    ).toBeVisible();
   }
 
   // We should see 5 video tiles everywhere now
@@ -101,13 +98,15 @@ widgetTest("Create and join a group call", async ({ addUser, browserName }) => {
   const florianFrame = florian.page
     .locator('iframe[title="Element Call"]')
     .contentFrame();
-  const florianMuteButton = florianFrame.getByTestId("incall_videomute");
-  await florianMuteButton.click();
+  const florianVideoButton = florianFrame.getByRole("switch", {
+    name: /video/,
+  });
+  await expect(florianVideoButton).toHaveAccessibleName("Stop video");
+  await expect(florianVideoButton).toBeChecked();
+  await florianVideoButton.click();
   // Now the button should indicate we can start video
-  await expect(florianMuteButton).toHaveAttribute(
-    "aria-label",
-    /^Start video$/,
-  );
+  await expect(florianVideoButton).toHaveAccessibleName("Start video");
+  await expect(florianVideoButton).not.toBeChecked();
 
   // wait a bit for the state to propagate
   await valere.page.waitForTimeout(3000);

playwright/widget/pip-call-button-interaction.test.ts

@@ -0,0 +1,73 @@
+/*
+Copyright 2026 Element Creations Ltd.
+
+SPDX-License-Identifier: AGPL-3.0-only OR LicenseRef-Element-Commercial
+Please see LICENSE in the repository root for full details.
+*/
+
+import { expect, test } from "@playwright/test";
+
+import { widgetTest } from "../fixtures/widget-user.ts";
+import { HOST1, TestHelpers } from "./test-helpers.ts";
+
+widgetTest("Footer interaction in PiP", async ({ addUser, browserName }) => {
+  test.skip(
+    browserName === "firefox",
+    "The is test is not working on firefox CI environment. No mic/audio device inputs so cam/mic are disabled",
+  );
+
+  test.slow();
+
+  const valere = await addUser("Valere", HOST1);
+
+  const callRoom = "CallRoom";
+  await TestHelpers.createRoom("CallRoom", valere.page);
+
+  await TestHelpers.createRoom("OtherRoom", valere.page);
+
+  await TestHelpers.switchToRoomNamed(valere.page, callRoom);
+
+  // Start the call as Valere
+  await TestHelpers.startCallInCurrentRoom(valere.page, false);
+  await expect(
+    valere.page.locator('iframe[title="Element Call"]'),
+  ).toBeVisible();
+
+  await TestHelpers.joinCallFromLobby(valere.page);
+  // wait a bit so that the PIP has rendered
+  await valere.page.waitForTimeout(600);
+
+  // Switch to the other room, the call should go to PIP
+  await TestHelpers.switchToRoomNamed(valere.page, "OtherRoom");
+
+  // We should see the PIP overlay
+  const iFrame = valere.page
+    .locator('iframe[title="Element Call"]')
+    .contentFrame();
+
+  {
+    // Check for a bug where the video had the wrong fit in PIP
+    const audioBtn = iFrame.getByRole("switch", { name: /microphone/ });
+    const videoBtn = iFrame.getByRole("switch", { name: /video/ });
+    await expect(
+      iFrame.getByRole("button", { name: "End call" }),
+    ).toBeVisible();
+    await expect(audioBtn).toBeVisible();
+    await expect(videoBtn).toBeVisible();
+    await expect(audioBtn).toHaveAccessibleName("Mute microphone");
+    await expect(audioBtn).toBeChecked();
+    await expect(videoBtn).toHaveAccessibleName("Stop video");
+    await expect(videoBtn).toBeChecked();
+
+    await videoBtn.click();
+    await audioBtn.click();
+
+    // stop hovering on any of the buttons
+    await iFrame.getByTestId("videoTile").hover();
+
+    await expect(audioBtn).toHaveAccessibleName("Unmute microphone");
+    await expect(audioBtn).toBeChecked();
+    await expect(videoBtn).toHaveAccessibleName("Start video");
+    await expect(videoBtn).not.toBeChecked();
+  }
+});

playwright/widget/pip-call.test.ts

@@ -0,0 +1,72 @@
+/*
+Copyright 2026 Element Creations Ltd.
+
+SPDX-License-Identifier: AGPL-3.0-only OR LicenseRef-Element-Commercial
+Please see LICENSE in the repository root for full details.
+*/
+
+import { expect, test } from "@playwright/test";
+
+import { widgetTest } from "../fixtures/widget-user.ts";
+import { HOST1, TestHelpers } from "./test-helpers.ts";
+
+widgetTest("Put call in PIP", async ({ addUser, browserName }) => {
+  test.skip(
+    browserName === "firefox",
+    "The is test is not working on firefox CI environment. No mic/audio device inputs so cam/mic are disabled",
+  );
+
+  test.slow();
+
+  const valere = await addUser("Valere", HOST1);
+  const timo = await addUser("Timo", HOST1);
+
+  const callRoom = "TeamRoom";
+  await TestHelpers.createRoom(callRoom, valere.page, [timo.mxId]);
+
+  await TestHelpers.createRoom("DoubleTask", valere.page);
+
+  await TestHelpers.acceptRoomInvite(callRoom, timo.page);
+
+  await TestHelpers.switchToRoomNamed(valere.page, callRoom);
+
+  // Start the call as Valere
+  await TestHelpers.startCallInCurrentRoom(valere.page, false);
+  await expect(
+    valere.page.locator('iframe[title="Element Call"]'),
+  ).toBeVisible();
+
+  await TestHelpers.joinCallFromLobby(valere.page);
+
+  await TestHelpers.joinCallInCurrentRoom(timo.page);
+
+  const frame = timo.page
+    .locator('iframe[title="Element Call"]')
+    .contentFrame();
+
+  // check that the video is on
+  await expect(
+    frame.getByRole("switch", { name: "Stop video", checked: true }),
+  ).toBeVisible();
+
+  // Switch to the other room, the call should go to PIP
+  await TestHelpers.switchToRoomNamed(valere.page, "DoubleTask");
+
+  // We should see the PIP overlay
+  await expect(valere.page.getByTestId("widget-pip-container")).toBeVisible();
+
+  {
+    // wait a bit so that the PIP has rendered the video
+    await valere.page.waitForTimeout(600);
+
+    // Check for a bug where the video had the wrong fit in PIP
+    const frame = valere.page
+      .locator('iframe[title="Element Call"]')
+      .contentFrame();
+    const videoElements = await frame.locator("video").all();
+    expect(videoElements.length).toBe(1);
+
+    const pipVideo = videoElements[0];
+    await expect(pipVideo).toHaveCSS("object-fit", "cover");
+  }
+});

playwright/widget/test-helpers.ts

@@ -276,4 +276,16 @@ export class TestHelpers {
       });
     }
   }
+
+  /**
+   * Switches to a room in the room list by its name.
+   * @param page - The EW page
+   * @param roomName - The name of the room to switch to
+   */
+  public static async switchToRoomNamed(
+    page: Page,
+    roomName: string,
+  ): Promise<void> {
+    await page.getByRole("option", { name: `Open room ${roomName}` }).click();
+  }
 }

playwright/widget/voice-call-dm.spec.ts

@@ -51,34 +51,36 @@ widgetTest(
       .contentFrame();
 
     // ASSERT the button states for whistler (the callee)
-    {
-      // The only way to know if it is muted or not is to look at the data-kind attribute..
-      const videoButton = whistlerFrame.getByTestId("incall_videomute");
-      // video should be off by default in a voice call
-      await expect(videoButton).toHaveAttribute("aria-label", /^Start video$/);
-
-      const audioButton = whistlerFrame.getByTestId("incall_mute");
-      // audio should be on for the voice call
-      await expect(audioButton).toHaveAttribute(
-        "aria-label",
-        /^Mute microphone$/,
-      );
-    }
+    // video should be off by default in a voice call
+    await expect(
+      whistlerFrame.getByRole("switch", {
+        name: "Start video",
+        checked: false,
+      }),
+    ).toBeVisible();
+    // audio should be on for the voice call
+    await expect(
+      whistlerFrame.getByRole("switch", {
+        name: "Mute microphone",
+        checked: true,
+      }),
+    ).toBeVisible();
 
     // ASSERT the button states for brools (the caller)
-    {
-      // The only way to know if it is muted or not is to look at the data-kind attribute..
-      const videoButton = brooksFrame.getByTestId("incall_videomute");
-      // video should be off by default in a voice call
-      await expect(videoButton).toHaveAttribute("aria-label", /^Start video$/);
-
-      const audioButton = brooksFrame.getByTestId("incall_mute");
-      // audio should be on for the voice call
-      await expect(audioButton).toHaveAttribute(
-        "aria-label",
-        /^Mute microphone$/,
-      );
-    }
+    // video should be off by default in a voice call
+    await expect(
+      whistlerFrame.getByRole("switch", {
+        name: "Start video",
+        checked: false,
+      }),
+    ).toBeVisible();
+    // audio should be on for the voice call
+    await expect(
+      whistlerFrame.getByRole("switch", {
+        name: "Mute microphone",
+        checked: true,
+      }),
+    ).toBeVisible();
 
     // In order to confirm that the call is disconnected we will check that the message composer is shown again.
     // So first we need to confirm that it is hidden when in the call.
@@ -90,10 +92,7 @@ widgetTest(
     ).not.toBeVisible();
 
     // ASSERT hanging up on one side ends the call for both
-    {
-      const hangupButton = brooksFrame.getByTestId("incall_leave");
-      await hangupButton.click();
-    }
+    await brooksFrame.getByRole("button", { name: "End call" }).click();
 
     // The widget should be closed on both sides and the timeline should be back on screen
     await expect(
@@ -142,34 +141,30 @@ widgetTest(
       .contentFrame();
 
     // ASSERT the button states for whistler (the callee)
-    {
-      // The only way to know if it is muted or not is to look at the data-kind attribute..
-      const videoButton = whistlerFrame.getByTestId("incall_videomute");
-      // video should be on by default in a voice call
-      await expect(videoButton).toHaveAttribute("aria-label", /^Stop video$/);
-
-      const audioButton = whistlerFrame.getByTestId("incall_mute");
-      // audio should be on for the voice call
-      await expect(audioButton).toHaveAttribute(
-        "aria-label",
-        /^Mute microphone$/,
-      );
-    }
+    // video should be off by default in a video call
+    await expect(
+      whistlerFrame.getByRole("switch", { name: "Stop video", checked: true }),
+    ).toBeVisible();
+    // audio should be on too
+    await expect(
+      whistlerFrame.getByRole("switch", {
+        name: "Mute microphone",
+        checked: true,
+      }),
+    ).toBeVisible();
 
     // ASSERT the button states for brools (the caller)
-    {
-      // The only way to know if it is muted or not is to look at the data-kind attribute..
-      const videoButton = brooksFrame.getByTestId("incall_videomute");
-      // video should be on by default in a voice call
-      await expect(videoButton).toHaveAttribute("aria-label", /^Stop video$/);
-
-      const audioButton = brooksFrame.getByTestId("incall_mute");
-      // audio should be on for the voice call
-      await expect(audioButton).toHaveAttribute(
-        "aria-label",
-        /^Mute microphone$/,
-      );
-    }
+    // video should be off by default in a video call
+    await expect(
+      whistlerFrame.getByRole("switch", { name: "Stop video", checked: true }),
+    ).toBeVisible();
+    // audio should be on too
+    await expect(
+      whistlerFrame.getByRole("switch", {
+        name: "Mute microphone",
+        checked: true,
+      }),
+    ).toBeVisible();
 
     // In order to confirm that the call is disconnected we will check that the message composer is shown again.
     // So first we need to confirm that it is hidden when in the call.
@@ -181,10 +176,7 @@ widgetTest(
     ).not.toBeVisible();
 
     // ASSERT hanging up on one side ends the call for both
-    {
-      const hangupButton = brooksFrame.getByTestId("incall_leave");
-      await hangupButton.click();
-    }
+    await brooksFrame.getByRole("button", { name: "End call" }).click();
 
     // The widget should be closed on both sides and the timeline should be back on screen
     await expect(

src/button/Button.tsx

@@ -22,23 +22,25 @@ import {
 import styles from "./Button.module.css";
 
 interface MicButtonProps extends ComponentPropsWithoutRef<"button"> {
-  muted: boolean;
+  enabled: boolean;
+  size?: "sm" | "lg";
 }
 
-export const MicButton: FC<MicButtonProps> = ({ muted, ...props }) => {
+export const MicButton: FC<MicButtonProps> = ({ enabled, ...props }) => {
   const { t } = useTranslation();
-  const Icon = muted ? MicOffSolidIcon : MicOnSolidIcon;
-  const label = muted
-    ? t("unmute_microphone_button_label")
-    : t("mute_microphone_button_label");
+  const Icon = enabled ? MicOnSolidIcon : MicOffSolidIcon;
+  const label = enabled
+    ? t("mute_microphone_button_label")
+    : t("unmute_microphone_button_label");
 
   return (
     <Tooltip label={label}>
       <CpdButton
         iconOnly
-        aria-label={label}
         Icon={Icon}
-        kind={muted ? "primary" : "secondary"}
+        kind={enabled ? "primary" : "secondary"}
+        role="switch"
+        aria-checked={enabled}
         {...props}
       />
     </Tooltip>
@@ -46,23 +48,25 @@ export const MicButton: FC<MicButtonProps> = ({ muted, ...props }) => {
 };
 
 interface VideoButtonProps extends ComponentPropsWithoutRef<"button"> {
-  muted: boolean;
+  enabled: boolean;
+  size?: "sm" | "lg";
 }
 
-export const VideoButton: FC<VideoButtonProps> = ({ muted, ...props }) => {
+export const VideoButton: FC<VideoButtonProps> = ({ enabled, ...props }) => {
   const { t } = useTranslation();
-  const Icon = muted ? VideoCallOffSolidIcon : VideoCallSolidIcon;
-  const label = muted
-    ? t("start_video_button_label")
-    : t("stop_video_button_label");
+  const Icon = enabled ? VideoCallSolidIcon : VideoCallOffSolidIcon;
+  const label = enabled
+    ? t("stop_video_button_label")
+    : t("start_video_button_label");
 
   return (
     <Tooltip label={label}>
       <CpdButton
         iconOnly
-        aria-label={label}
         Icon={Icon}
-        kind={muted ? "primary" : "secondary"}
+        kind={enabled ? "primary" : "secondary"}
+        role="switch"
+        aria-checked={enabled}
         {...props}
       />
     </Tooltip>
@@ -71,6 +75,7 @@ export const VideoButton: FC<VideoButtonProps> = ({ muted, ...props }) => {
 
 interface ShareScreenButtonProps extends ComponentPropsWithoutRef<"button"> {
   enabled: boolean;
+  size: "sm" | "lg";
 }
 
 export const ShareScreenButton: FC<ShareScreenButtonProps> = ({
@@ -88,13 +93,19 @@ export const ShareScreenButton: FC<ShareScreenButtonProps> = ({
         iconOnly
         Icon={ShareScreenSolidIcon}
         kind={enabled ? "primary" : "secondary"}
+        role="switch"
+        aria-checked={enabled}
         {...props}
       />
     </Tooltip>
   );
 };
 
-export const EndCallButton: FC<ComponentPropsWithoutRef<"button">> = ({
+interface EndCallButtonProps extends ComponentPropsWithoutRef<"button"> {
+  size?: "sm" | "lg";
+}
+
+export const EndCallButton: FC<EndCallButtonProps> = ({
   className,
   ...props
 }) => {
@@ -105,7 +116,6 @@ export const EndCallButton: FC<ComponentPropsWithoutRef<"button">> = ({
       <CpdButton
         className={classNames(className, styles.endCall)}
         iconOnly
-        aria-label={t("hangup_button_label")}
         Icon={EndCallIcon}
         destructive
         {...props}
@@ -114,9 +124,10 @@ export const EndCallButton: FC<ComponentPropsWithoutRef<"button">> = ({
   );
 };
 
-export const SettingsButton: FC<ComponentPropsWithoutRef<"button">> = (
-  props,
-) => {
+interface SettingsButtonProps extends ComponentPropsWithoutRef<"button"> {
+  size?: "sm" | "lg";
+}
+export const SettingsButton: FC<SettingsButtonProps> = (props) => {
   const { t } = useTranslation();
 
   return (

src/button/ReactionToggleButton.tsx

@@ -166,6 +166,7 @@ export function ReactionPopupMenu({
 interface ReactionToggleButtonProps extends ComponentPropsWithoutRef<"button"> {
   identifier: string;
   vm: CallViewModel;
+  size?: "sm" | "lg";
 }
 
 export function ReactionToggleButton({

src/grid/TileWrapper.tsx

@@ -27,7 +27,13 @@ interface Props<M, R extends HTMLElement> {
       state: Parameters<Handler<"drag", EventTypes["drag"]>>[0],
     ) => void
   > | null;
+  /**
+   * The width this tile will have once its animations have settled.
+   */
   targetWidth: number;
+  /**
+   * The width this tile will have once its animations have settled.
+   */
   targetHeight: number;
   model: M;
   Tile: ComponentType<TileProps<M, R>>;

src/livekit/MatrixAudioRenderer.test.tsx

@@ -30,7 +30,13 @@ import {
 } from "../utils/test";
 import { initializeWidget } from "../widget";
 initializeWidget();
-export const TestAudioContextConstructor = vi.fn(() => testAudioContext);
+export const TestAudioContextConstructor = vi.fn(
+  class {
+    public constructor() {
+      return testAudioContext;
+    }
+  },
+);
 
 const MediaDevicesProvider = MediaDevicesContext.MediaDevicesContext.Provider;
 

src/room/InCallView.module.css

@@ -65,6 +65,7 @@ Please see LICENSE in the repository root for full details.
 .footer.overlay.hidden {
   display: grid;
   opacity: 0;
+  pointer-events: none;
 }
 
 .footer.overlay:has(:focus-visible) {
@@ -107,22 +108,9 @@ Please see LICENSE in the repository root for full details.
   }
 }
 
-@media (max-width: 370px) {
-  .shareScreen {
-    display: none;
-  }
-
-  @media (max-height: 400px) {
-    .footer {
-      display: none;
-    }
-  }
-}
-
-@media (max-width: 320px) {
-  .invite,
-  .raiseHand {
-    display: none;
+@media (max-height: 800px) {
+  .footer {
+    padding-block: var(--cpd-space-8x);
   }
 }
 
@@ -132,9 +120,27 @@ Please see LICENSE in the repository root for full details.
   }
 }
 
-@media (max-height: 800px) {
-  .footer {
-    padding-block: var(--cpd-space-8x);
+@media (max-width: 370px) {
+  .shareScreen {
+    display: none;
+  }
+
+  /* PIP custom css */
+  @media (max-height: 400px) {
+    .shareScreen {
+      display: flex;
+    }
+    .footer {
+      padding-block-start: var(--cpd-space-3x);
+      padding-block-end: var(--cpd-space-2x);
+    }
+  }
+}
+
+@media (max-width: 320px) {
+  .invite,
+  .raiseHand {
+    display: none;
   }
 }
 

src/room/InCallView.tsx

@@ -9,8 +9,8 @@ import { IconButton, Text, Tooltip } from "@vector-im/compound-web";
 import { type MatrixClient, type Room as MatrixRoom } from "matrix-js-sdk";
 import {
   type FC,
-  type PointerEvent,
-  type TouchEvent,
+  type MouseEvent as ReactMouseEvent,
+  type PointerEvent as ReactPointerEvent,
   useCallback,
   useEffect,
   useMemo,
@@ -110,8 +110,6 @@ import { ObservableScope } from "../state/ObservableScope.ts";
 
 const logger = rootLogger.getChild("[InCallView]");
 
-const maxTapDurationMs = 400;
-
 export interface ActiveCallProps extends Omit<
   InCallViewProps,
   "vm" | "livekitRoom" | "connState"
@@ -334,40 +332,20 @@ export const InCallView: FC<InCallViewProps> = ({
     ) : null;
   }, [ringOverlay]);
 
-  // Ideally we could detect taps by listening for click events and checking
-  // that the pointerType of the event is "touch", but this isn't yet supported
-  // in Safari: https://developer.mozilla.org/en-US/docs/Web/API/Element/click_event#browser_compatibility
-  // Instead we have to watch for sufficiently fast touch events.
-  const touchStart = useRef<number | null>(null);
-  const onTouchStart = useCallback(() => (touchStart.current = Date.now()), []);
-  const onTouchEnd = useCallback(() => {
-    const start = touchStart.current;
-    if (start !== null && Date.now() - start <= maxTapDurationMs)
-      vm.tapScreen();
-    touchStart.current = null;
-  }, [vm]);
-  const onTouchCancel = useCallback(() => (touchStart.current = null), []);
-
-  // We also need to tell the footer controls to prevent touch events from
-  // bubbling up, or else the footer will be dismissed before a click/change
-  // event can be registered on the control
-  const onControlsTouchEnd = useCallback(
-    (e: TouchEvent) => {
-      // Somehow applying pointer-events: none to the controls when the footer
-      // is hidden is not enough to stop clicks from happening as the footer
-      // becomes visible, so we check manually whether the footer is shown
-      if (showFooter) {
-        e.stopPropagation();
-        vm.tapControls();
-      } else {
-        e.preventDefault();
-      }
+  const onViewClick = useCallback(
+    (e: ReactMouseEvent) => {
+      if (
+        (e.nativeEvent as PointerEvent).pointerType === "touch" &&
+        // If an interactive element was tapped, don't count this as a tap on the screen
+        (e.target as Element).closest?.("button, input") === null
+      )
+        vm.tapScreen();
     },
-    [vm, showFooter],
+    [vm],
   );
 
   const onPointerMove = useCallback(
-    (e: PointerEvent) => {
+    (e: ReactPointerEvent) => {
       if (e.pointerType === "mouse") vm.hoverScreen();
     },
     [vm],
@@ -606,8 +584,8 @@ export const InCallView: FC<InCallViewProps> = ({
           vm={layout.spotlight}
           expanded
           onToggleExpanded={null}
-          targetWidth={gridBounds.height}
-          targetHeight={gridBounds.width}
+          targetWidth={gridBounds.width}
+          targetHeight={gridBounds.height}
           showIndicators={false}
           focusable={!contentObscured}
           aria-hidden={contentObscured}
@@ -662,20 +640,21 @@ export const InCallView: FC<InCallViewProps> = ({
 
   const buttons: JSX.Element[] = [];
 
+  const buttonSize = layout.type === "pip" ? "sm" : "lg";
   buttons.push(
     <MicButton
+      size={buttonSize}
       key="audio"
-      muted={!audioEnabled}
+      enabled={audioEnabled}
       onClick={toggleAudio ?? undefined}
-      onTouchEnd={onControlsTouchEnd}
       disabled={toggleAudio === null}
       data-testid="incall_mute"
     />,
     <VideoButton
+      size={buttonSize}
       key="video"
-      muted={!videoEnabled}
+      enabled={videoEnabled}
       onClick={toggleVideo ?? undefined}
-      onTouchEnd={onControlsTouchEnd}
       disabled={toggleVideo === null}
       data-testid="incall_videomute"
     />,
@@ -683,11 +662,11 @@ export const InCallView: FC<InCallViewProps> = ({
   if (vm.toggleScreenSharing !== null) {
     buttons.push(
       <ShareScreenButton
+        size={buttonSize}
         key="share_screen"
         className={styles.shareScreen}
         enabled={sharingScreen}
         onClick={vm.toggleScreenSharing}
-        onTouchEnd={onControlsTouchEnd}
         data-testid="incall_screenshare"
       />,
     );
@@ -695,30 +674,30 @@ export const InCallView: FC<InCallViewProps> = ({
   if (supportsReactions) {
     buttons.push(
       <ReactionToggleButton
+        size={buttonSize}
         vm={vm}
         key="raise_hand"
         className={styles.raiseHand}
         identifier={`${client.getUserId()}:${client.getDeviceId()}`}
-        onTouchEnd={onControlsTouchEnd}
       />,
     );
   }
   if (layout.type !== "pip")
     buttons.push(
       <SettingsButton
+        size={buttonSize}
         key="settings"
         onClick={openSettings}
-        onTouchEnd={onControlsTouchEnd}
       />,
     );
 
   buttons.push(
     <EndCallButton
+      size={buttonSize}
       key="end_call"
       onClick={function (): void {
         vm.hangup();
       }}
-      onTouchEnd={onControlsTouchEnd}
       data-testid="incall_leave"
     />,
   );
@@ -751,7 +730,6 @@ export const InCallView: FC<InCallViewProps> = ({
           className={styles.layout}
           layout={gridMode}
           setLayout={setGridMode}
-          onTouchEnd={onControlsTouchEnd}
         />
       )}
     </div>
@@ -760,12 +738,13 @@ export const InCallView: FC<InCallViewProps> = ({
   const allConnections = useBehavior(vm.allConnections$);
 
   return (
+    // The onClick handler here exists to control the visibility of the footer,
+    // and the footer is also viewable by moving focus into it, so this is fine.
+    // eslint-disable-next-line jsx-a11y/no-static-element-interactions, jsx-a11y/click-events-have-key-events
     <div
       className={styles.inRoom}
       ref={containerRef}
-      onTouchStart={onTouchStart}
-      onTouchEnd={onTouchEnd}
-      onTouchCancel={onTouchCancel}
+      onClick={onViewClick}
       onPointerMove={onPointerMove}
       onPointerOut={onPointerOut}
     >

src/room/LayoutToggle.tsx

@@ -5,7 +5,7 @@ SPDX-License-Identifier: AGPL-3.0-only OR LicenseRef-Element-Commercial
 Please see LICENSE in the repository root for full details.
 */
 
-import { type ChangeEvent, type FC, type TouchEvent, useCallback } from "react";
+import { type ChangeEvent, type FC, useCallback } from "react";
 import { useTranslation } from "react-i18next";
 import { Tooltip } from "@vector-im/compound-web";
 import {
@@ -22,15 +22,9 @@ interface Props {
   layout: Layout;
   setLayout: (layout: Layout) => void;
   className?: string;
-  onTouchEnd?: (e: TouchEvent) => void;
 }
 
-export const LayoutToggle: FC<Props> = ({
-  layout,
-  setLayout,
-  className,
-  onTouchEnd,
-}) => {
+export const LayoutToggle: FC<Props> = ({ layout, setLayout, className }) => {
   const { t } = useTranslation();
 
   const onChange = useCallback(
@@ -47,7 +41,6 @@ export const LayoutToggle: FC<Props> = ({
           value="spotlight"
           checked={layout === "spotlight"}
           onChange={onChange}
-          onTouchEnd={onTouchEnd}
         />
       </Tooltip>
       <SpotlightIcon aria-hidden width={24} height={24} />
@@ -58,7 +51,6 @@ export const LayoutToggle: FC<Props> = ({
           value="grid"
           checked={layout === "grid"}
           onChange={onChange}
-          onTouchEnd={onTouchEnd}
         />
       </Tooltip>
       <GridIcon aria-hidden width={24} height={24} />

src/room/LobbyView.tsx

@@ -230,12 +230,12 @@ export const LobbyView: FC<Props> = ({
           {recentsButtonInFooter && recentsButton}
           <div className={inCallStyles.buttons}>
             <MicButton
-              muted={!audioEnabled}
+              enabled={audioEnabled}
               onClick={toggleAudio ?? undefined}
               disabled={toggleAudio === null}
             />
             <VideoButton
-              muted={!videoEnabled}
+              enabled={videoEnabled}
               onClick={toggleVideo ?? undefined}
               disabled={toggleVideo === null}
             />

src/room/__snapshots__/InCallView.test.tsx.snap

@@ -285,14 +285,14 @@ exports[`InCallView > rendering > renders 1`] = `
         class="buttons"
       >
         <button
+          aria-checked="false"
           aria-disabled="true"
-          aria-label="Unmute microphone"
           aria-labelledby="_r_8_"
           class="_button_13vu4_8 _has-icon_13vu4_60 _icon-only_13vu4_53"
-          data-kind="primary"
+          data-kind="secondary"
           data-size="lg"
           data-testid="incall_mute"
-          role="button"
+          role="switch"
           tabindex="0"
         >
           <svg
@@ -309,14 +309,14 @@ exports[`InCallView > rendering > renders 1`] = `
           </svg>
         </button>
         <button
+          aria-checked="false"
           aria-disabled="true"
-          aria-label="Start video"
           aria-labelledby="_r_d_"
           class="_button_13vu4_8 _has-icon_13vu4_60 _icon-only_13vu4_53"
-          data-kind="primary"
+          data-kind="secondary"
           data-size="lg"
           data-testid="incall_videomute"
-          role="button"
+          role="switch"
           tabindex="0"
         >
           <svg
@@ -354,7 +354,6 @@ exports[`InCallView > rendering > renders 1`] = `
           </svg>
         </button>
         <button
-          aria-label="End call"
           aria-labelledby="_r_n_"
           class="_button_13vu4_8 endCall _has-icon_13vu4_60 _icon-only_13vu4_53 _destructive_13vu4_110"
           data-kind="primary"

src/settings/DeveloperSettingsTab.test.tsx

@@ -5,13 +5,16 @@ SPDX-License-Identifier: AGPL-3.0-only OR LicenseRef-Element-Commercial
 Please see LICENSE in the repository root for full details.
 */
 
-import { describe, expect, it, vi } from "vitest";
-import { render, waitFor } from "@testing-library/react";
-import { type Room as LivekitRoom } from "livekit-client";
+import { afterEach, describe, expect, it, type Mock, vi } from "vitest";
+import { render, waitFor, screen } from "@testing-library/react";
+import userEvent from "@testing-library/user-event";
+import { TooltipProvider } from "@vector-im/compound-web";
 
 import type { MatrixClient } from "matrix-js-sdk";
+import type { Room as LivekitRoom } from "livekit-client";
 import { DeveloperSettingsTab } from "./DeveloperSettingsTab";
-
+import { getSFUConfigWithOpenID } from "../livekit/openIDSFU";
+import { customLivekitUrl as customLivekitUrlSetting } from "./settings";
 // Mock url params hook to avoid environment-dependent snapshot churn.
 vi.mock("../UrlParams", () => ({
   useUrlParams: (): { mocked: boolean; answer: number } => ({
@@ -20,6 +23,14 @@ vi.mock("../UrlParams", () => ({
   }),
 }));
 
+// IMPORTANT: mock the same specifier used by DeveloperSettingsTab
+vi.mock("../livekit/openIDSFU", () => ({
+  getSFUConfigWithOpenID: vi.fn().mockResolvedValue({
+    url: "mock-url",
+    jwt: "mock-jwt",
+  }),
+}));
+
 // Provide a minimal mock of a Livekit Room structure used by the component.
 function createMockLivekitRoom(
   wsUrl: string,
@@ -86,6 +97,7 @@ describe("DeveloperSettingsTab", () => {
     const { container } = render(
       <DeveloperSettingsTab
         client={client}
+        roomId={"#room:example.org"}
         livekitRooms={livekitRooms}
         env={{ MY_MOCK_ENV: 10, ENV: "test" } as unknown as ImportMetaEnv}
       />,
@@ -99,4 +111,141 @@ describe("DeveloperSettingsTab", () => {
 
     expect(container).toMatchSnapshot();
   });
+  describe("custom livekit url", () => {
+    afterEach(() => {
+      customLivekitUrlSetting.setValue(null);
+    });
+    const client = {
+      doesServerSupportUnstableFeature: vi.fn().mockResolvedValue(true),
+      getCrypto: () => ({ getVersion: (): string => "x" }),
+      getUserId: () => "@u:hs",
+      getDeviceId: () => "DEVICE",
+    } as unknown as MatrixClient;
+    it("will not update custom livekit url without roomId", async () => {
+      const user = userEvent.setup();
+
+      render(
+        <TooltipProvider>
+          <DeveloperSettingsTab
+            client={client}
+            env={{} as unknown as ImportMetaEnv}
+          />
+        </TooltipProvider>,
+      );
+
+      const input = screen.getByLabelText("Custom Livekit-url");
+      await user.clear(input);
+      await user.type(input, "wss://example.livekit.invalid");
+
+      const saveButton = screen.getByRole("button", { name: "Save" });
+      await user.click(saveButton);
+      expect(getSFUConfigWithOpenID).not.toHaveBeenCalled();
+
+      expect(customLivekitUrlSetting.getValue()).toBe(null);
+    });
+    it("will not update custom livekit url without text in input", async () => {
+      const user = userEvent.setup();
+
+      render(
+        <TooltipProvider>
+          <DeveloperSettingsTab
+            client={client}
+            roomId="#testRoom"
+            env={{} as unknown as ImportMetaEnv}
+          />
+        </TooltipProvider>,
+      );
+
+      const input = screen.getByLabelText("Custom Livekit-url");
+      await user.clear(input);
+
+      const saveButton = screen.getByRole("button", { name: "Save" });
+      await user.click(saveButton);
+      expect(getSFUConfigWithOpenID).not.toHaveBeenCalled();
+
+      expect(customLivekitUrlSetting.getValue()).toBe(null);
+    });
+    it("will not update custom livekit url when pressing cancel", async () => {
+      const user = userEvent.setup();
+
+      render(
+        <TooltipProvider>
+          <DeveloperSettingsTab
+            client={client}
+            roomId="#testRoom"
+            env={{} as unknown as ImportMetaEnv}
+          />
+        </TooltipProvider>,
+      );
+
+      const input = screen.getByLabelText("Custom Livekit-url");
+      await user.clear(input);
+      await user.type(input, "wss://example.livekit.invalid");
+
+      const cancelButton = screen.getByRole("button", {
+        name: "Reset overwrite",
+      });
+      await user.click(cancelButton);
+      expect(getSFUConfigWithOpenID).not.toHaveBeenCalled();
+
+      expect(customLivekitUrlSetting.getValue()).toBe(null);
+    });
+    it("will update custom livekit url", async () => {
+      const user = userEvent.setup();
+
+      render(
+        <TooltipProvider>
+          <DeveloperSettingsTab
+            client={client}
+            roomId="#testRoom"
+            env={{} as unknown as ImportMetaEnv}
+          />
+        </TooltipProvider>,
+      );
+
+      const input = screen.getByLabelText("Custom Livekit-url");
+      await user.clear(input);
+      await user.type(input, "wss://example.livekit.valid");
+
+      const saveButton = screen.getByRole("button", { name: "Save" });
+      await user.click(saveButton);
+      expect(getSFUConfigWithOpenID).toHaveBeenCalledWith(
+        expect.anything(),
+        expect.anything(),
+        "wss://example.livekit.valid",
+        "#testRoom",
+      );
+
+      expect(customLivekitUrlSetting.getValue()).toBe(
+        "wss://example.livekit.valid",
+      );
+    });
+    it("will show error on invalid url", async () => {
+      const user = userEvent.setup();
+
+      render(
+        <TooltipProvider>
+          <DeveloperSettingsTab
+            client={client}
+            roomId="#testRoom"
+            env={{} as unknown as ImportMetaEnv}
+          />
+        </TooltipProvider>,
+      );
+
+      const input = screen.getByLabelText("Custom Livekit-url");
+      await user.clear(input);
+      await user.type(input, "wss://example.livekit.valid");
+
+      const saveButton = screen.getByRole("button", { name: "Save" });
+      (getSFUConfigWithOpenID as Mock).mockImplementation(() => {
+        throw new Error("Invalid URL");
+      });
+      await user.click(saveButton);
+      expect(
+        screen.getByText("invalid URL (did not update)"),
+      ).toBeInTheDocument();
+      expect(customLivekitUrlSetting.getValue()).toBe(null);
+    });
+  });
 });

src/settings/DeveloperSettingsTab.tsx

@@ -22,6 +22,7 @@ import {
 import { logger } from "matrix-js-sdk/lib/logger";
 import {
   EditInPlace,
+  ErrorMessage,
   Root as Form,
   Heading,
   HelpMessage,
@@ -45,9 +46,11 @@ import {
 } from "./settings";
 import styles from "./DeveloperSettingsTab.module.css";
 import { useUrlParams } from "../UrlParams";
+import { getSFUConfigWithOpenID } from "../livekit/openIDSFU";
 
 interface Props {
   client: MatrixClient;
+  roomId?: string;
   livekitRooms?: {
     room: LivekitRoom;
     url: string;
@@ -60,6 +63,7 @@ interface Props {
 export const DeveloperSettingsTab: FC<Props> = ({
   client,
   livekitRooms,
+  roomId,
   env,
 }) => {
   const { t } = useTranslation();
@@ -97,6 +101,8 @@ export const DeveloperSettingsTab: FC<Props> = ({
     alwaysShowIphoneEarpieceSetting,
   );
 
+  const [customLivekitUrlUpdateError, setCustomLivekitUrlUpdateError] =
+    useState<string | null>(null);
   const [customLivekitUrl, setCustomLivekitUrl] = useSetting(
     customLivekitUrlSetting,
   );
@@ -234,14 +240,36 @@ export const DeveloperSettingsTab: FC<Props> = ({
         savingLabel={t("developer_mode.custom_livekit_url.saving")}
         cancelButtonLabel={t("developer_mode.custom_livekit_url.reset")}
         onSave={useCallback(
-          (e: React.FormEvent<HTMLFormElement>) => {
-            setCustomLivekitUrl(
-              customLivekitUrlTextBuffer === ""
-                ? null
-                : customLivekitUrlTextBuffer,
-            );
+          async (e: React.FormEvent<HTMLFormElement>): Promise<void> => {
+            if (
+              roomId === undefined ||
+              customLivekitUrlTextBuffer === "" ||
+              customLivekitUrlTextBuffer === null
+            ) {
+              setCustomLivekitUrl(null);
+              return;
+            }
+
+            try {
+              const userId = client.getUserId();
+              const deviceId = client.getDeviceId();
+
+              if (userId === null || deviceId === null) {
+                throw new Error("Invalid user or device ID");
+              }
+              await getSFUConfigWithOpenID(
+                client,
+                { userId, deviceId, memberId: "" },
+                customLivekitUrlTextBuffer,
+                roomId,
+              );
+              setCustomLivekitUrlUpdateError(null);
+              setCustomLivekitUrl(customLivekitUrlTextBuffer);
+            } catch {
+              setCustomLivekitUrlUpdateError("invalid URL (did not update)");
+            }
           },
-          [setCustomLivekitUrl, customLivekitUrlTextBuffer],
+          [customLivekitUrlTextBuffer, setCustomLivekitUrl, client, roomId],
         )}
         value={customLivekitUrlTextBuffer ?? ""}
         onChange={useCallback(
@@ -256,7 +284,12 @@ export const DeveloperSettingsTab: FC<Props> = ({
           },
           [setCustomLivekitUrl],
         )}
-      />
+        serverInvalid={customLivekitUrlUpdateError !== null}
+      >
+        {customLivekitUrlUpdateError !== null && (
+          <ErrorMessage>{customLivekitUrlUpdateError}</ErrorMessage>
+        )}
+      </EditInPlace>
       <Heading as="h3" type="body" weight="semibold" size="lg">
         {t("developer_mode.matrixRTCMode.title")}
       </Heading>

src/settings/SettingsModal.tsx

@@ -213,6 +213,7 @@ export const SettingsModal: FC<Props> = ({
         env={import.meta.env}
         client={client}
         livekitRooms={livekitRooms}
+        roomId={roomId}
       />
     ),
   };

src/settings/settings.ts

@@ -34,15 +34,20 @@ export class Setting<T> {
 
     this._value$ = new BehaviorSubject(initialValue);
     this.value$ = this._value$;
+    this._lastUpdateReason$ = new BehaviorSubject<string | null>(null);
+    this.lastUpdateReason$ = this._lastUpdateReason$;
   }
 
   private readonly key: string;
 
   private readonly _value$: BehaviorSubject<T>;
+  private readonly _lastUpdateReason$: BehaviorSubject<string | null>;
   public readonly value$: Behavior<T>;
+  public readonly lastUpdateReason$: Behavior<string | null>;
 
-  public readonly setValue = (value: T): void => {
+  public readonly setValue = (value: T, reason?: string): void => {
     this._value$.next(value);
+    this._lastUpdateReason$.next(reason ?? null);
     localStorage.setItem(this.key, JSON.stringify(value));
   };
   public readonly getValue = (): T => {

src/state/CallViewModel/CallViewModel.ts

@@ -63,7 +63,7 @@ import {
   playReactionsSound,
   showReactions,
 } from "../../settings/settings";
-import { isFirefox } from "../../Platform";
+import { isFirefox, platform } from "../../Platform";
 import { setPipEnabled$ } from "../../controls";
 import { TileStore } from "../TileStore";
 import { gridLikeLayout } from "../GridLikeLayout";
@@ -1271,7 +1271,7 @@ export function createCallViewModel$(
       switchMap((mode) => {
         switch (mode) {
           case "pip":
-            return of(false);
+            return of(platform === "desktop" ? true : false);
           case "normal":
           case "narrow":
             return of(true);

src/state/PipLayout.ts

@@ -5,6 +5,7 @@ SPDX-License-Identifier: AGPL-3.0-only OR LicenseRef-Element-Commercial
 Please see LICENSE in the repository root for full details.
 */
 
+import { platform } from "../Platform.ts";
 import { type PipLayout, type PipLayoutMedia } from "./layout-types.ts";
 import { type TileStore } from "./TileStore";
 
@@ -16,7 +17,11 @@ export function pipLayout(
   prevTiles: TileStore,
 ): [PipLayout, TileStore] {
   const update = prevTiles.from(0);
-  update.registerSpotlight(media.spotlight, true);
+  // Dont maximise in pip on EW since we want the rounded corners and the footer
+  update.registerSpotlight(
+    media.spotlight,
+    platform === "desktop" ? false : true,
+  );
   const tiles = update.build();
   return [
     {

src/state/media/MediaViewModel.test.ts

@@ -9,6 +9,7 @@ import { expect, onTestFinished, test, vi } from "vitest";
 import {
   type LocalTrackPublication,
   LocalVideoTrack,
+  Track,
   TrackEvent,
 } from "livekit-client";
 import { waitFor } from "@testing-library/dom";
@@ -21,6 +22,7 @@ import {
   mockRemoteMedia,
   withTestScheduler,
   mockRemoteParticipant,
+  mockRemoteScreenShare,
 } from "../../utils/test";
 import { constant } from "../Behavior";
 
@@ -91,17 +93,69 @@ test("control a participant's volume", () => {
   });
 });
 
-test("toggle fit/contain for a participant's video", () => {
-  const vm = mockRemoteMedia(rtcMembership, {}, mockRemoteParticipant({}));
+test("control a participant's screen share volume", () => {
+  const setVolumeSpy = vi.fn();
+  const vm = mockRemoteScreenShare(
+    rtcMembership,
+    {},
+    mockRemoteParticipant({ setVolume: setVolumeSpy }),
+  );
   withTestScheduler(({ expectObservable, schedule }) => {
-    schedule("-ab|", {
-      a: () => vm.toggleCropVideo(),
-      b: () => vm.toggleCropVideo(),
+    schedule("-ab---c---d|", {
+      a() {
+        // Try muting by toggling
+        vm.togglePlaybackMuted();
+        expect(setVolumeSpy).toHaveBeenLastCalledWith(
+          0,
+          Track.Source.ScreenShareAudio,
+        );
+      },
+      b() {
+        // Try unmuting by dragging the slider back up
+        vm.adjustPlaybackVolume(0.6);
+        vm.adjustPlaybackVolume(0.8);
+        vm.commitPlaybackVolume();
+        expect(setVolumeSpy).toHaveBeenCalledWith(
+          0.6,
+          Track.Source.ScreenShareAudio,
+        );
+        expect(setVolumeSpy).toHaveBeenLastCalledWith(
+          0.8,
+          Track.Source.ScreenShareAudio,
+        );
+      },
+      c() {
+        // Try muting by dragging the slider back down
+        vm.adjustPlaybackVolume(0.2);
+        vm.adjustPlaybackVolume(0);
+        vm.commitPlaybackVolume();
+        expect(setVolumeSpy).toHaveBeenCalledWith(
+          0.2,
+          Track.Source.ScreenShareAudio,
+        );
+        expect(setVolumeSpy).toHaveBeenLastCalledWith(
+          0,
+          Track.Source.ScreenShareAudio,
+        );
+      },
+      d() {
+        // Try unmuting by toggling
+        vm.togglePlaybackMuted();
+        // The volume should return to the last non-zero committed volume
+        expect(setVolumeSpy).toHaveBeenLastCalledWith(
+          0.8,
+          Track.Source.ScreenShareAudio,
+        );
+      },
     });
-    expectObservable(vm.cropVideo$).toBe("abc", {
-      a: true,
-      b: false,
-      c: true,
+    expectObservable(vm.playbackVolume$).toBe("ab(cd)(ef)g", {
+      a: 1,
+      b: 0,
+      c: 0.6,
+      d: 0.8,
+      e: 0.2,
+      f: 0,
+      g: 0.8,
     });
   });
 });

src/state/media/RemoteScreenShareViewModel.ts

@@ -6,8 +6,8 @@ SPDX-License-Identifier: AGPL-3.0-only OR LicenseRef-Element-Commercial
 Please see LICENSE in the repository root for full details.
 */
 
-import { type RemoteParticipant } from "livekit-client";
-import { map } from "rxjs";
+import { Track, type RemoteParticipant } from "livekit-client";
+import { map, of, switchMap } from "rxjs";
 
 import { type Behavior } from "../Behavior";
 import {
@@ -16,13 +16,20 @@ import {
   createBaseScreenShare,
 } from "./ScreenShareViewModel";
 import { type ObservableScope } from "../ObservableScope";
+import { createVolumeControls, type VolumeControls } from "../VolumeControls";
+import { observeTrackReference$ } from "../observeTrackReference";
 
-export interface RemoteScreenShareViewModel extends BaseScreenShareViewModel {
+export interface RemoteScreenShareViewModel
+  extends BaseScreenShareViewModel, VolumeControls {
   local: false;
   /**
    * Whether this screen share's video should be displayed.
    */
   videoEnabled$: Behavior<boolean>;
+  /**
+   * Whether this screen share should be considered to have an audio track.
+   */
+  audioEnabled$: Behavior<boolean>;
 }
 
 export interface RemoteScreenShareInputs extends BaseScreenShareInputs {
@@ -36,9 +43,30 @@ export function createRemoteScreenShare(
 ): RemoteScreenShareViewModel {
   return {
     ...createBaseScreenShare(scope, inputs),
+    ...createVolumeControls(scope, {
+      pretendToBeDisconnected$,
+      sink$: scope.behavior(
+        inputs.participant$.pipe(
+          map(
+            (p) => (volume) =>
+              p?.setVolume(volume, Track.Source.ScreenShareAudio),
+          ),
+        ),
+      ),
+    }),
     local: false,
     videoEnabled$: scope.behavior(
       pretendToBeDisconnected$.pipe(map((disconnected) => !disconnected)),
     ),
+    audioEnabled$: scope.behavior(
+      inputs.participant$.pipe(
+        switchMap((p) =>
+          p
+            ? observeTrackReference$(p, Track.Source.ScreenShareAudio)
+            : of(null),
+        ),
+        map(Boolean),
+      ),
+    ),
   };
 }

src/state/media/UserMediaViewModel.ts

@@ -7,6 +7,7 @@ Please see LICENSE in the repository root for full details.
 */
 
 import {
+  BehaviorSubject,
   combineLatest,
   map,
   type Observable,
@@ -30,9 +31,9 @@ import {
 } from "./MemberMediaViewModel";
 import { type RemoteUserMediaViewModel } from "./RemoteUserMediaViewModel";
 import { type ObservableScope } from "../ObservableScope";
-import { createToggle$ } from "../../utils/observable";
 import { showConnectionStats } from "../../settings/settings";
 import { observeRtpStreamStats$ } from "./observeRtpStreamStats";
+import { videoFit$, videoSizeFromParticipant$ } from "../../utils/videoFit.ts";
 
 /**
  * A participant's user media (i.e. their microphone and camera feed).
@@ -46,7 +47,7 @@ export interface BaseUserMediaViewModel extends MemberMediaViewModel {
   speaking$: Behavior<boolean>;
   audioEnabled$: Behavior<boolean>;
   videoEnabled$: Behavior<boolean>;
-  cropVideo$: Behavior<boolean>;
+  videoFit$: Behavior<"cover" | "contain">;
   toggleCropVideo: () => void;
   /**
    * The expected identity of the LiveKit participant. Exposed for debugging.
@@ -60,6 +61,13 @@ export interface BaseUserMediaViewModel extends MemberMediaViewModel {
   videoStreamStats$: Observable<
     RTCInboundRtpStreamStats | RTCOutboundRtpStreamStats | undefined
   >;
+  /**
+   * Set the target dimensions of the HTML element (final dimension after anim).
+   * This can be used to determine the best video fit (fit to frame / keep ratio).
+   * @param targetWidth - The target width of the HTML element displaying the video.
+   * @param targetHeight - The target height of the HTML element displaying the video.
+   */
+  setTargetDimensions: (targetWidth: number, targetHeight: number) => void;
 }
 
 export interface BaseUserMediaInputs extends Omit<
@@ -90,6 +98,12 @@ export function createBaseUserMedia(
   );
   const toggleCropVideo$ = new Subject<void>();
 
+  // The target size of the video element, used to determine the best video fit.
+  // The target size is the final size of the HTML element after any animations have completed.
+  const targetSize$ = new BehaviorSubject<
+    { width: number; height: number } | undefined
+  >(undefined);
+
   return {
     ...createMemberMedia(scope, {
       ...inputs,
@@ -115,7 +129,11 @@ export function createBaseUserMedia(
     videoEnabled$: scope.behavior(
       media$.pipe(map((m) => m?.cameraTrack?.isMuted === false)),
     ),
-    cropVideo$: createToggle$(scope, true, toggleCropVideo$),
+    videoFit$: videoFit$(
+      scope,
+      videoSizeFromParticipant$(participant$),
+      targetSize$,
+    ),
     toggleCropVideo: () => toggleCropVideo$.next(),
     rtcBackendIdentity,
     handRaised$,
@@ -139,5 +157,8 @@ export function createBaseUserMedia(
         return observeRtpStreamStats$(p, Track.Source.Camera, statsType);
       }),
     ),
+    setTargetDimensions: (targetWidth: number, targetHeight: number): void => {
+      targetSize$.next({ width: targetWidth, height: targetHeight });
+    },
   };
 }

src/state/media/observeRtpStreamStats.ts

@@ -67,3 +67,12 @@ export function observeInboundRtpStreamStats$(
     map((x) => x as RTCInboundRtpStreamStats | undefined),
   );
 }
+
+export function observeOutboundRtpStreamStats$(
+  participant: Participant,
+  source: Track.Source,
+): Observable<RTCOutboundRtpStreamStats | undefined> {
+  return observeRtpStreamStats$(participant, source, "outbound-rtp").pipe(
+    map((x) => x as RTCOutboundRtpStreamStats | undefined),
+  );
+}

src/tile/GridTile.tsx

@@ -11,6 +11,7 @@ import {
   type ReactNode,
   type Ref,
   useCallback,
+  useEffect,
   useRef,
   useState,
 } from "react";
@@ -26,7 +27,6 @@ import {
   VolumeOffIcon,
   VisibilityOnIcon,
   UserProfileIcon,
-  ExpandIcon,
   VolumeOffSolidIcon,
   SwitchCameraSolidIcon,
 } from "@vector-im/compound-design-tokens/assets/web/icons";
@@ -87,6 +87,8 @@ const UserMediaTile: FC<UserMediaTileProps> = ({
   displayName,
   mxcAvatarUrl,
   focusable,
+  targetWidth,
+  targetHeight,
   ...props
 }) => {
   const { toggleRaisedHand } = useReactionsSender();
@@ -103,18 +105,19 @@ const UserMediaTile: FC<UserMediaTileProps> = ({
   const audioEnabled = useBehavior(vm.audioEnabled$);
   const videoEnabled = useBehavior(vm.videoEnabled$);
   const speaking = useBehavior(vm.speaking$);
-  const cropVideo = useBehavior(vm.cropVideo$);
-  const onSelectFitContain = useCallback(
-    (e: Event) => {
-      e.preventDefault();
-      vm.toggleCropVideo();
-    },
-    [vm],
-  );
+  const videoFit = useBehavior(vm.videoFit$);
+
   const rtcBackendIdentity = vm.rtcBackendIdentity;
   const handRaised = useBehavior(vm.handRaised$);
   const reaction = useBehavior(vm.reaction$);
 
+  // Whenever bounds change, inform the viewModel
+  useEffect(() => {
+    if (targetWidth > 0 && targetHeight > 0) {
+      vm.setTargetDimensions(targetWidth, targetHeight);
+    }
+  }, [targetWidth, targetHeight, vm]);
+
   const AudioIcon = playbackMuted
     ? VolumeOffSolidIcon
     : audioEnabled
@@ -130,12 +133,10 @@ const UserMediaTile: FC<UserMediaTileProps> = ({
   const menu = (
     <>
       {menuStart}
-      <ToggleMenuItem
-        Icon={ExpandIcon}
-        label={t("video_tile.change_fit_contain")}
-        checked={cropVideo}
-        onSelect={onSelectFitContain}
-      />
+      {/*
+       No additional menu item (used to be the manual fit to frame.
+       Placeholder for future menu items that should be placed here.
+       */}
       {menuEnd}
     </>
   );
@@ -154,7 +155,7 @@ const UserMediaTile: FC<UserMediaTileProps> = ({
       unencryptedWarning={unencryptedWarning}
       encryptionStatus={encryptionStatus}
       videoEnabled={videoEnabled}
-      videoFit={cropVideo ? "cover" : "contain"}
+      videoFit={videoFit}
       className={classNames(className, styles.tile, {
         [styles.speaking]: showSpeaking,
         [styles.handRaised]: !showSpeaking && handRaised,
@@ -200,6 +201,8 @@ const UserMediaTile: FC<UserMediaTileProps> = ({
       audioStreamStats={audioStreamStats}
       videoStreamStats={videoStreamStats}
       rtcBackendIdentity={rtcBackendIdentity}
+      targetWidth={targetWidth}
+      targetHeight={targetHeight}
       {...props}
     />
   );

src/tile/SpotlightTile.module.css

@@ -84,7 +84,6 @@ Please see LICENSE in the repository root for full details.
 .expand {
   appearance: none;
   cursor: pointer;
-  opacity: 0;
   padding: var(--cpd-space-2x);
   border: none;
   border-radius: var(--cpd-radius-pill-effect);
@@ -108,6 +107,35 @@ Please see LICENSE in the repository root for full details.
   z-index: 1;
 }
 
+.volumeSlider {
+  width: 100%;
+  min-width: 172px;
+}
+
+/* Disable the hover effect for the screen share volume menu button */
+.volumeMenuItem:hover {
+  background: transparent;
+  cursor: default;
+}
+
+.volumeMenuItem {
+  gap: var(--cpd-space-3x);
+}
+
+.menuMuteButton {
+  appearance: none;
+  background: none;
+  border: none;
+  padding: 0;
+  cursor: pointer;
+  display: flex;
+}
+
+/* Make icons change color with the theme */
+.menuMuteButton > svg {
+  color: var(--cpd-color-icon-primary);
+}
+
 .expand > svg {
   display: block;
   color: var(--cpd-color-icon-primary);
@@ -119,17 +147,22 @@ Please see LICENSE in the repository root for full details.
   }
 }
 
-.expand:active {
+.expand:active,
+.expand[data-state="open"] {
   background: var(--cpd-color-gray-100);
 }
 
 @media (hover) {
+  .tile > div > button {
+    opacity: 0;
+  }
   .tile:hover > div > button {
     opacity: 1;
   }
 }
 
-.tile:has(:focus-visible) > div > button {
+.tile:has(:focus-visible) > div > button,
+.tile > div:has([data-state="open"]) > button {
   opacity: 1;
 }
 

src/tile/SpotlightTile.test.tsx

@@ -9,6 +9,7 @@ import { test, expect, vi } from "vitest";
 import { isInaccessible, render, screen } from "@testing-library/react";
 import { axe } from "vitest-axe";
 import userEvent from "@testing-library/user-event";
+import { TooltipProvider } from "@vector-im/compound-web";
 
 import { SpotlightTile } from "./SpotlightTile";
 import {
@@ -18,6 +19,7 @@ import {
   mockLocalMedia,
   mockRemoteMedia,
   mockRemoteParticipant,
+  mockRemoteScreenShare,
 } from "../utils/test";
 import { SpotlightTileViewModel } from "../state/TileViewModel";
 import { constant } from "../state/Behavior";
@@ -78,3 +80,63 @@ test("SpotlightTile is accessible", async () => {
   await user.click(screen.getByRole("button", { name: "Expand" }));
   expect(toggleExpanded).toHaveBeenCalled();
 });
+
+test("Screen share volume UI is shown when screen share has audio", async () => {
+  const vm = mockRemoteScreenShare(
+    mockRtcMembership("@alice:example.org", "AAAA"),
+    {},
+    mockRemoteParticipant({}),
+  );
+
+  vi.spyOn(vm, "audioEnabled$", "get").mockReturnValue(constant(true));
+
+  const toggleExpanded = vi.fn();
+  const { container } = render(
+    <TooltipProvider>
+      <SpotlightTile
+        vm={new SpotlightTileViewModel(constant([vm]), constant(false))}
+        targetWidth={300}
+        targetHeight={200}
+        expanded={false}
+        onToggleExpanded={toggleExpanded}
+        showIndicators
+        focusable
+      />
+    </TooltipProvider>,
+  );
+
+  expect(await axe(container)).toHaveNoViolations();
+
+  // Volume menu button should exist
+  expect(screen.queryByRole("button", { name: /volume/i })).toBeInTheDocument();
+});
+
+test("Screen share volume UI is hidden when screen share has no audio", async () => {
+  const vm = mockRemoteScreenShare(
+    mockRtcMembership("@alice:example.org", "AAAA"),
+    {},
+    mockRemoteParticipant({}),
+  );
+
+  vi.spyOn(vm, "audioEnabled$", "get").mockReturnValue(constant(false));
+
+  const toggleExpanded = vi.fn();
+  const { container } = render(
+    <SpotlightTile
+      vm={new SpotlightTileViewModel(constant([vm]), constant(false))}
+      targetWidth={300}
+      targetHeight={200}
+      expanded={false}
+      onToggleExpanded={toggleExpanded}
+      showIndicators
+      focusable
+    />,
+  );
+
+  expect(await axe(container)).toHaveNoViolations();
+
+  // Volume menu button should not exist
+  expect(
+    screen.queryByRole("button", { name: /volume/i }),
+  ).not.toBeInTheDocument();
+});

src/tile/SpotlightTile.tsx

@@ -20,6 +20,10 @@ import {
   CollapseIcon,
   ChevronLeftIcon,
   ChevronRightIcon,
+  VolumeOffIcon,
+  VolumeOnIcon,
+  VolumeOffSolidIcon,
+  VolumeOnSolidIcon,
 } from "@vector-im/compound-design-tokens/assets/web/icons";
 import { animated } from "@react-spring/web";
 import { type Observable, map } from "rxjs";
@@ -27,6 +31,7 @@ import { useObservableRef } from "observable-hooks";
 import { useTranslation } from "react-i18next";
 import classNames from "classnames";
 import { type TrackReferenceOrPlaceholder } from "@livekit/components-core";
+import { Menu, MenuItem } from "@vector-im/compound-web";
 
 import FullScreenMaximiseIcon from "../icons/FullScreenMaximise.svg?react";
 import FullScreenMinimiseIcon from "../icons/FullScreenMinimise.svg?react";
@@ -45,6 +50,8 @@ import { type UserMediaViewModel } from "../state/media/UserMediaViewModel";
 import { type ScreenShareViewModel } from "../state/media/ScreenShareViewModel";
 import { type RemoteScreenShareViewModel } from "../state/media/RemoteScreenShareViewModel";
 import { type MediaViewModel } from "../state/media/MediaViewModel";
+import { Slider } from "../Slider";
+import { platform } from "../Platform";
 
 interface SpotlightItemBaseProps {
   ref?: Ref<HTMLDivElement>;
@@ -104,12 +111,12 @@ const SpotlightUserMediaItem: FC<SpotlightUserMediaItemProps> = ({
   vm,
   ...props
 }) => {
-  const cropVideo = useBehavior(vm.cropVideo$);
+  const videoFit = useBehavior(vm.videoFit$);
   const videoEnabled = useBehavior(vm.videoEnabled$);
 
   const baseProps: SpotlightUserMediaItemBaseProps &
     RefAttributes<HTMLDivElement> = {
-    videoFit: cropVideo ? "cover" : "contain",
+    videoFit,
     videoEnabled,
     ...props,
   };
@@ -151,7 +158,13 @@ const SpotlightRemoteScreenShareItem: FC<
 interface SpotlightItemProps {
   ref?: Ref<HTMLDivElement>;
   vm: MediaViewModel;
+  /**
+   * The width this tile will have once its animations have settled.
+   */
   targetWidth: number;
+  /**
+   * The height this tile will have once its animations have settled.
+   */
   targetHeight: number;
   focusable: boolean;
   intersectionObserver$: Observable<IntersectionObserver>;
@@ -173,6 +186,16 @@ const SpotlightItem: FC<SpotlightItemProps> = ({
   "aria-hidden": ariaHidden,
 }) => {
   const ourRef = useRef<HTMLDivElement | null>(null);
+
+  // Whenever target bounds change, inform the viewModel
+  useEffect(() => {
+    if (targetWidth > 0 && targetHeight > 0) {
+      if (vm.type != "screen share") {
+        vm.setTargetDimensions(targetWidth, targetHeight);
+      }
+    }
+  }, [targetWidth, targetHeight, vm]);
+
   const ref = useMergedRefs(ourRef, theirRef);
   const focusUrl = useBehavior(vm.focusUrl$);
   const displayName = useBehavior(vm.displayName$);
@@ -224,6 +247,73 @@ const SpotlightItem: FC<SpotlightItemProps> = ({
 
 SpotlightItem.displayName = "SpotlightItem";
 
+interface ScreenShareVolumeButtonProps {
+  vm: RemoteScreenShareViewModel;
+}
+
+const ScreenShareVolumeButton: FC<ScreenShareVolumeButtonProps> = ({ vm }) => {
+  const { t } = useTranslation();
+
+  const audioEnabled = useBehavior(vm.audioEnabled$);
+  const playbackMuted = useBehavior(vm.playbackMuted$);
+  const playbackVolume = useBehavior(vm.playbackVolume$);
+
+  const VolumeIcon = playbackMuted ? VolumeOffIcon : VolumeOnIcon;
+  const VolumeSolidIcon = playbackMuted
+    ? VolumeOffSolidIcon
+    : VolumeOnSolidIcon;
+
+  const [volumeMenuOpen, setVolumeMenuOpen] = useState(false);
+  const onMuteButtonClick = useCallback(() => vm.togglePlaybackMuted(), [vm]);
+  const onVolumeChange = useCallback(
+    (v: number) => vm.adjustPlaybackVolume(v),
+    [vm],
+  );
+  const onVolumeCommit = useCallback(() => vm.commitPlaybackVolume(), [vm]);
+
+  return (
+    audioEnabled && (
+      <Menu
+        open={volumeMenuOpen}
+        onOpenChange={setVolumeMenuOpen}
+        title={t("video_tile.screen_share_volume")}
+        side="top"
+        align="end"
+        trigger={
+          <button
+            className={styles.expand}
+            aria-label={t("video_tile.screen_share_volume")}
+          >
+            <VolumeSolidIcon aria-hidden width={20} height={20} />
+          </button>
+        }
+      >
+        <MenuItem
+          as="div"
+          className={styles.volumeMenuItem}
+          onSelect={null}
+          label={null}
+          hideChevron={true}
+        >
+          <button className={styles.menuMuteButton} onClick={onMuteButtonClick}>
+            <VolumeIcon aria-hidden width={24} height={24} />
+          </button>
+          <Slider
+            className={styles.volumeSlider}
+            label={t("video_tile.volume")}
+            value={playbackVolume}
+            min={0}
+            max={1}
+            step={0.01}
+            onValueChange={onVolumeChange}
+            onValueCommit={onVolumeCommit}
+          />
+        </MenuItem>
+      </Menu>
+    )
+  );
+};
+
 interface Props {
   ref?: Ref<HTMLDivElement>;
   vm: SpotlightTileViewModel;
@@ -258,6 +348,7 @@ export const SpotlightTile: FC<Props> = ({
   const latestMedia = useLatest(media);
   const latestVisibleId = useLatest(visibleId);
   const visibleIndex = media.findIndex((vm) => vm.id === visibleId);
+  const visibleMedia = media.at(visibleIndex);
   const canGoBack = visibleIndex > 0;
   const canGoToNext = visibleIndex !== -1 && visibleIndex < media.length - 1;
 
@@ -365,16 +456,21 @@ export const SpotlightTile: FC<Props> = ({
           />
         ))}
       </div>
-      <div className={styles.bottomRightButtons}>
-        <button
-          className={classNames(styles.expand)}
-          aria-label={"maximise"}
-          onClick={onToggleFullscreen}
-          tabIndex={focusable ? undefined : -1}
-        >
-          <FullScreenIcon aria-hidden width={20} height={20} />
-        </button>
 
+      <div className={styles.bottomRightButtons}>
+        {visibleMedia?.type === "screen share" && !visibleMedia.local && (
+          <ScreenShareVolumeButton vm={visibleMedia} />
+        )}
+        {platform === "desktop" && (
+          <button
+            className={classNames(styles.expand)}
+            aria-label={"maximise"}
+            onClick={onToggleFullscreen}
+            tabIndex={focusable ? undefined : -1}
+          >
+            <FullScreenIcon aria-hidden width={20} height={20} />
+          </button>
+        )}
         {onToggleExpanded && (
           <button
             className={classNames(styles.expand)}

src/useAudioContext.test.tsx

@@ -90,11 +90,18 @@ export const testAudioContext = {
   createStereoPanner: vi.fn().mockReturnValue(panNode),
   close: vi.fn().mockResolvedValue(undefined),
 };
-export const TestAudioContextConstructor = vi.fn(() => testAudioContext);
+
+const TestAudioContext = vi.fn(
+  class {
+    public constructor() {
+      return testAudioContext;
+    }
+  },
+);
 
 let user: UserEvent;
 beforeEach(() => {
-  vi.stubGlobal("AudioContext", TestAudioContextConstructor);
+  vi.stubGlobal("AudioContext", TestAudioContext);
   user = userEvent.setup();
 });
 

src/utils/test.ts

@@ -70,6 +70,10 @@ import {
   createRemoteUserMedia,
   type RemoteUserMediaViewModel,
 } from "../state/media/RemoteUserMediaViewModel";
+import {
+  createRemoteScreenShare,
+  type RemoteScreenShareViewModel,
+} from "../state/media/RemoteScreenShareViewModel";
 
 export function withFakeTimers(continuation: () => void): void {
   vi.useFakeTimers();
@@ -393,6 +397,31 @@ export function mockRemoteMedia(
   });
 }
 
+export function mockRemoteScreenShare(
+  rtcMember: CallMembership,
+  roomMember: Partial<RoomMember>,
+  participant: RemoteParticipant | null,
+  livekitRoom: LivekitRoom | undefined = mockLivekitRoom(
+    {},
+    {
+      remoteParticipants$: of(participant ? [participant] : []),
+    },
+  ),
+): RemoteScreenShareViewModel {
+  const member = mockMatrixRoomMember(rtcMember, roomMember);
+  return createRemoteScreenShare(testScope(), {
+    id: "screenshare",
+    userId: member.userId,
+    participant$: constant(participant),
+    encryptionSystem: { kind: E2eeType.PER_PARTICIPANT },
+    livekitRoom$: constant(livekitRoom),
+    focusUrl$: constant("https://rtc-example.org"),
+    pretendToBeDisconnected$: constant(false),
+    displayName$: constant(member.rawDisplayName ?? "nodisplayname"),
+    mxcAvatarUrl$: constant(member.getMxcAvatarUrl()),
+  });
+}
+
 export function mockConfig(
   config: Partial<ResolvedConfigOptions> = {},
 ): MockInstance<() => ResolvedConfigOptions> {
@@ -414,10 +443,10 @@ export class MockRTCSession extends TypedEventEmitter<
   public asMockedSession(): MockedObject<MatrixRTCSession> {
     const session = this as unknown as MockedObject<MatrixRTCSession>;
 
-    vi.mocked(session).reemitEncryptionKeys = vi
+    session.reemitEncryptionKeys = vi
       .fn<() => void>()
       .mockReturnValue(undefined);
-    vi.mocked(session).getOldestMembership = vi
+    session.getOldestMembership = vi
       .fn<() => CallMembership | undefined>()
       .mockReturnValue(this.memberships[0]);
 

src/utils/videoFit.test.ts

@@ -0,0 +1,263 @@
+/*
+Copyright 2026 Element Creations Ltd.
+
+SPDX-License-Identifier: AGPL-3.0-only OR LicenseRef-Element-Commercial
+Please see LICENSE in the repository root for full details.
+*/
+
+import { describe, expect, test, vi } from "vitest";
+import {
+  LocalTrack,
+  type LocalTrackPublication,
+  type RemoteTrackPublication,
+  Track,
+} from "livekit-client";
+
+import { ObservableScope } from "../state/ObservableScope";
+import { videoFit$, videoSizeFromParticipant$ } from "./videoFit";
+import { constant } from "../state/Behavior";
+import {
+  flushPromises,
+  mockLocalParticipant,
+  mockRemoteParticipant,
+} from "./test";
+
+describe("videoFit$ defaults", () => {
+  test.each([
+    {
+      videoSize: { width: 1920, height: 1080 },
+      tileSize: undefined,
+    },
+    {
+      videoSize: { width: 1080, height: 1920 },
+      tileSize: undefined,
+    },
+    {
+      videoSize: undefined,
+      tileSize: { width: 1920, height: 1080 },
+    },
+    {
+      videoSize: undefined,
+      tileSize: { width: 1080, height: 1920 },
+    },
+  ])(
+    "videoFit$ returns `cover` when videoSize is $videoSize and tileSize is $tileSize",
+    ({ videoSize, tileSize }) => {
+      const scope = new ObservableScope();
+      const videoSize$ = constant(videoSize);
+      const tileSize$ = constant(tileSize);
+
+      const fit = videoFit$(scope, videoSize$, tileSize$);
+      expect(fit.value).toBe("cover");
+    },
+  );
+});
+
+const VIDEO_480_L = { width: 640, height: 480 };
+const VIDEO_720_L = { width: 1280, height: 720 };
+const VIDEO_1080_L = { width: 1920, height: 1080 };
+
+// Some sizes from real world testing, which don't match the standard video sizes exactly
+const TILE_SIZE_1_L = { width: 180, height: 135 };
+const TILE_SIZE_3_P = { width: 379, height: 542 };
+const TILE_SIZE_4_L = { width: 957, height: 542 };
+// This is the size of an iPhone Xr in portrait mode
+const TILE_SIZE_5_P = { width: 414, height: 896 };
+
+export function invertSize(size: { width: number; height: number }): {
+  width: number;
+  height: number;
+} {
+  return {
+    width: size.height,
+    height: size.width,
+  };
+}
+
+test.each([
+  {
+    videoSize: VIDEO_480_L,
+    tileSize: TILE_SIZE_1_L,
+    expected: "cover",
+  },
+  {
+    videoSize: invertSize(VIDEO_480_L),
+    tileSize: TILE_SIZE_1_L,
+    expected: "contain",
+  },
+  {
+    videoSize: VIDEO_720_L,
+    tileSize: TILE_SIZE_4_L,
+    expected: "cover",
+  },
+  {
+    videoSize: invertSize(VIDEO_720_L),
+    tileSize: TILE_SIZE_4_L,
+    expected: "contain",
+  },
+  {
+    videoSize: invertSize(VIDEO_1080_L),
+    tileSize: TILE_SIZE_3_P,
+    expected: "cover",
+  },
+  {
+    videoSize: VIDEO_1080_L,
+    tileSize: TILE_SIZE_5_P,
+    expected: "contain",
+  },
+  {
+    videoSize: invertSize(VIDEO_1080_L),
+    tileSize: TILE_SIZE_5_P,
+    expected: "cover",
+  },
+  {
+    // square video
+    videoSize: { width: 400, height: 400 },
+    tileSize: VIDEO_480_L,
+    expected: "contain",
+  },
+  {
+    // Should default to cover if the initial size is 0:0.
+    // Or else it will cause a flash of "contain" mode until the real size is loaded, which can be jarring.
+    videoSize: VIDEO_480_L,
+    tileSize: { width: 0, height: 0 },
+    expected: "cover",
+  },
+  {
+    videoSize: { width: 0, height: 0 },
+    tileSize: VIDEO_480_L,
+    expected: "cover",
+  },
+])(
+  "videoFit$ returns $expected when videoSize is $videoSize and tileSize is $tileSize",
+  ({ videoSize, tileSize, expected }) => {
+    const scope = new ObservableScope();
+    const videoSize$ = constant(videoSize);
+    const tileSize$ = constant(tileSize);
+
+    const fit = videoFit$(scope, videoSize$, tileSize$);
+    expect(fit.value).toBe(expected);
+  },
+);
+
+describe("extracting video size from participant stats", () => {
+  function createMockRtpStats(
+    isInbound: boolean,
+    props: Partial<RTCInboundRtpStreamStats | RTCOutboundRtpStreamStats> = {},
+  ): RTCInboundRtpStreamStats | RTCOutboundRtpStreamStats {
+    const baseStats = {
+      id: "mock-stats-id",
+      timestamp: Date.now(),
+      type: isInbound ? "inbound-rtp" : "outbound-rtp",
+      kind: "video",
+      ...props,
+    };
+
+    return baseStats as RTCInboundRtpStreamStats | RTCOutboundRtpStreamStats;
+  }
+
+  test("get stats for local user", async () => {
+    const localParticipant = mockLocalParticipant({
+      identity: "@local:example.org:AAAAAA",
+    });
+
+    const mockReport: RTCStatsReport = new Map([
+      [
+        "OT01V639885149",
+        createMockRtpStats(false, {
+          frameWidth: 1280,
+          frameHeight: 720,
+        }),
+      ],
+    ]);
+
+    const track = {
+      source: Track.Source.Camera,
+      getRTCStatsReport: vi
+        .fn()
+        .mockImplementation(async () => Promise.resolve(mockReport)),
+    } as Partial<LocalTrack> as LocalTrack;
+
+    // Set up the prototype chain (there is an instanceof check in getRTCStatsReport)
+    Object.setPrototypeOf(track, LocalTrack.prototype);
+
+    localParticipant.getTrackPublication = vi
+      .fn()
+      .mockImplementation((source: Track.Source) => {
+        if (source === Track.Source.Camera) {
+          return {
+            track,
+          } as unknown as LocalTrackPublication;
+        } else {
+          return undefined;
+        }
+      });
+
+    const videoDimensions$ = videoSizeFromParticipant$(
+      constant(localParticipant),
+    );
+
+    const publishedDimensions: { width: number; height: number }[] = [];
+    videoDimensions$.subscribe((dimensions) => {
+      if (dimensions) publishedDimensions.push(dimensions);
+    });
+
+    await flushPromises();
+
+    const dimension = publishedDimensions.pop();
+    expect(dimension).toEqual({ width: 1280, height: 720 });
+  });
+
+  test("get stats for remote user", async () => {
+    // vi.useFakeTimers()
+    const remoteParticipant = mockRemoteParticipant({
+      identity: "@bob:example.org:AAAAAA",
+    });
+
+    const mockReport: RTCStatsReport = new Map([
+      [
+        "OT01V639885149",
+        createMockRtpStats(true, {
+          frameWidth: 480,
+          frameHeight: 640,
+        }),
+      ],
+    ]);
+
+    const track = {
+      source: Track.Source.Camera,
+      getRTCStatsReport: vi
+        .fn()
+        .mockImplementation(async () => Promise.resolve(mockReport)),
+    } as Partial<LocalTrack> as LocalTrack;
+
+    // Set up the prototype chain (there is an instanceof check in getRTCStatsReport)
+    Object.setPrototypeOf(track, LocalTrack.prototype);
+
+    remoteParticipant.getTrackPublication = vi
+      .fn()
+      .mockImplementation((source: Track.Source) => {
+        if (source === Track.Source.Camera) {
+          return {
+            track,
+          } as unknown as RemoteTrackPublication;
+        } else {
+          return undefined;
+        }
+      });
+
+    const videoDimensions$ = videoSizeFromParticipant$(
+      constant(remoteParticipant),
+    );
+
+    const publishedDimensions: { width: number; height: number }[] = [];
+    videoDimensions$.subscribe((dimensions) => {
+      if (dimensions) publishedDimensions.push(dimensions);
+    });
+
+    await flushPromises();
+
+    const dimension = publishedDimensions.pop();
+    expect(dimension).toEqual({ width: 480, height: 640 });
+  });
+});

src/utils/videoFit.ts

@@ -0,0 +1,111 @@
+/*
+Copyright 2026 Element Creations Ltd.
+
+SPDX-License-Identifier: AGPL-3.0-only OR LicenseRef-Element-Commercial
+Please see LICENSE in the repository root for full details.
+*/
+
+import { combineLatest, map, type Observable, of, switchMap } from "rxjs";
+import {
+  type LocalParticipant,
+  type RemoteParticipant,
+  Track,
+} from "livekit-client";
+
+import { type ObservableScope } from "../state/ObservableScope.ts";
+import { type Behavior } from "../state/Behavior.ts";
+import {
+  observeInboundRtpStreamStats$,
+  observeOutboundRtpStreamStats$,
+} from "../state/media/observeRtpStreamStats";
+
+type Size = {
+  width: number;
+  height: number;
+};
+
+/**
+ * Computes the appropriate video fit mode ("cover" or "contain") based on the aspect ratios of the video and the tile.
+ * - If the video and tile have the same orientation (both landscape or both portrait), we use "cover" to fill the tile, even if it means cropping.
+ * - If the video and tile have different orientations, we use "contain" to ensure the entire video is visible, even if it means letterboxing (black bars).
+ * @param scope - the ObservableScope to create the Behavior in
+ * @param videoSize$ - an Observable of the video size (width and height) or undefined if the size is not yet known (no data yet received).
+ * @param tileSize$ - an Observable of the tile size (width and height) or undefined if the size is not yet known (not yet rendered).
+ */
+export function videoFit$(
+  scope: ObservableScope,
+  videoSize$: Observable<Size | undefined>,
+  tileSize$: Observable<Size | undefined>,
+): Behavior<"cover" | "contain"> {
+  const fit$ = combineLatest([videoSize$, tileSize$]).pipe(
+    map(([videoSize, tileSize]) => {
+      if (!videoSize || !tileSize) {
+        // If we don't have the sizes, default to cover to avoid black bars.
+        // This is a reasonable default as it will ensure the video fills the tile, even if it means cropping.
+        return "cover";
+      }
+      if (
+        videoSize.width === 0 ||
+        videoSize.height === 0 ||
+        tileSize.width === 0 ||
+        tileSize.height === 0
+      ) {
+        // If we have invalid sizes (e.g. width or height is 0), default to cover to avoid black bars.
+        return "cover";
+      }
+      const videoAspectRatio = videoSize.width / videoSize.height;
+      const tileAspectRatio = tileSize.width / tileSize.height;
+
+      // If video is landscape (ratio > 1) and tile is portrait (ratio < 1) or vice versa,
+      // we want to use "contain" (fit) mode to avoid excessive cropping
+      const videoIsLandscape = videoAspectRatio > 1;
+      const tileIsLandscape = tileAspectRatio > 1;
+
+      // If the orientations are the same, use the cover mode (Preserves the aspect ratio, and the image fills the container.)
+      // If they're not the same orientation, use the contain mode (Preserves the aspect ratio, but the image is letterboxed - black bars- to fit within the container.)
+      return videoIsLandscape === tileIsLandscape ? "cover" : "contain";
+    }),
+  );
+
+  return scope.behavior(fit$, "cover");
+}
+
+/**
+ * Helper function to get the video size from a participant.
+ * It observes the participant's video track stats and extracts the frame width and height.
+ * @param participant$ - an Observable of a LocalParticipant or RemoteParticipant, or null if no participant is selected.
+ * @returns an Observable of the video size (width and height) or undefined if the size cannot be determined.
+ */
+export function videoSizeFromParticipant$(
+  participant$: Observable<LocalParticipant | RemoteParticipant | null>,
+): Observable<{ width: number; height: number } | undefined> {
+  return participant$
+    .pipe(
+      // If we have a participant, observe their video track stats. If not, return undefined.
+      switchMap((p) => {
+        if (!p) return of(undefined);
+        if (p.isLocal) {
+          return observeOutboundRtpStreamStats$(p, Track.Source.Camera);
+        } else {
+          return observeInboundRtpStreamStats$(p, Track.Source.Camera);
+        }
+      }),
+    )
+    .pipe(
+      // Extract the frame width and height from the stats. If we don't have valid stats, return undefined.
+      map((stats) => {
+        if (!stats) return undefined;
+        if (
+          // For video tracks, frameWidth and frameHeight should be numbers. If they're not, we can't determine the size.
+          typeof stats.frameWidth !== "number" ||
+          typeof stats.frameHeight !== "number"
+        ) {
+          return undefined;
+        }
+        return {
+          width: stats.frameWidth,
+          height: stats.frameHeight,
+        };
+      }),
+    );
+}