Add sounds for ringing (#3490)

* add wait for pickup overlay

Signed-off-by: Timo K <toger5@hotmail.de>

* refactor and leave logic

Signed-off-by: Timo K <toger5@hotmail.de>

* recursive play sound logic

Signed-off-by: Timo K <toger5@hotmail.de>

* review

Signed-off-by: Timo K <toger5@hotmail.de>

* text color

Signed-off-by: Timo K <toger5@hotmail.de>

* overlay styling and interval fixes

Signed-off-by: Timo K <toger5@hotmail.de>

* fix permissions and styling

Signed-off-by: Timo K <toger5@hotmail.de>

* fix always getting pickup sound

Signed-off-by: Timo K <toger5@hotmail.de>

* Add sound effects for declined,timeout and ringtone

* better ringtone

* Integrate sounds

* Ensure leave sound does not play

* Remove unused blocked sound

* fix test

* Improve tests

* Loop ring sound inside Audio context for better perf.

* lint

* better ringtone

* Update to delay ringtone logic.

* lint + fix test

* Tidy up ring sync and add comments.

* lint

* Refactor onLeave to take a sound so we don't need to repeat the sound

* fix import

---------

Signed-off-by: Timo K <toger5@hotmail.de>
Co-authored-by: Timo K <toger5@hotmail.de>

src/room/CallEventAudioRenderer.test.tsx

@@ -54,6 +54,8 @@ beforeEach(() => {
   playSound = vitest.fn();
   (useAudioContext as MockedFunction<typeof useAudioContext>).mockReturnValue({
     playSound,
+    playSoundLooping: vitest.fn(),
+    soundDuration: {},
   });
 });
 
@@ -105,6 +107,20 @@ test("plays a sound when a user leaves", () => {
   expect(playSound).toBeCalledWith("left");
 });
 
+test("does not play a sound before the call is successful", () => {
+  const { vm, rtcMemberships$ } = getBasicCallViewModelEnvironment(
+    [local, alice],
+    [localRtcMember],
+    { waitForCallPickup: true },
+  );
+  render(<CallEventAudioRenderer vm={vm} />);
+
+  act(() => {
+    rtcMemberships$.next([localRtcMember]);
+  });
+  expect(playSound).not.toBeCalledWith("left");
+});
+
 test("plays no sound when the participant list is more than the maximum size", () => {
   const mockRtcMemberships: CallMembership[] = [localRtcMember];
   for (let i = 0; i < MAX_PARTICIPANT_COUNT_FOR_SOUND; i++) {

src/room/CallEventAudioRenderer.tsx

@@ -16,6 +16,10 @@ import handSoundOgg from "../sound/raise_hand.ogg";
 import handSoundMp3 from "../sound/raise_hand.mp3";
 import screenShareStartedOgg from "../sound/screen_share_started.ogg";
 import screenShareStartedMp3 from "../sound/screen_share_started.mp3";
+import declineMp3 from "../sound/call_declined.mp3?url";
+import declineOgg from "../sound/call_declined.ogg?url";
+import timeoutMp3 from "../sound/call_timeout.mp3?url";
+import timeoutOgg from "../sound/call_timeout.ogg?url";
 import { useAudioContext } from "../useAudioContext";
 import { prefetchSounds } from "../soundUtils";
 import { useLatest } from "../useLatest";
@@ -37,8 +41,18 @@ export const callEventAudioSounds = prefetchSounds({
     mp3: screenShareStartedMp3,
     ogg: screenShareStartedOgg,
   },
+  decline: {
+    mp3: declineMp3,
+    ogg: declineOgg,
+  },
+  timeout: {
+    mp3: timeoutMp3,
+    ogg: timeoutOgg,
+  },
 });
 
+export type CallEventSounds = keyof Awaited<typeof callEventAudioSounds>;
+
 export function CallEventAudioRenderer({
   vm,
   muted,

src/room/GroupCallView.test.tsx

@@ -12,6 +12,7 @@ import {
   onTestFinished,
   test,
   vi,
+  vitest,
 } from "vitest";
 import { render, waitFor, screen, act } from "@testing-library/react";
 import { type MatrixClient, JoinRule, type RoomState } from "matrix-js-sdk";
@@ -97,13 +98,15 @@ beforeEach(() => {
   playSound = vi.fn();
   (useAudioContext as MockedFunction<typeof useAudioContext>).mockReturnValue({
     playSound,
+    playSoundLooping: vi.fn(),
+    soundDuration: {},
   });
   // A trivial implementation of Active call to ensure we are testing GroupCallView exclusively here.
   (ActiveCall as MockedFunction<typeof ActiveCall>).mockImplementation(
     ({ onLeave }) => {
       return (
         <div>
-          <button onClick={() => onLeave()}>Leave</button>
+          <button onClick={() => onLeave("user")}>Leave</button>
         </div>
       );
     },
@@ -209,6 +212,8 @@ test("GroupCallView plays a leave sound synchronously in widget mode", async ()
     );
   (useAudioContext as MockedFunction<typeof useAudioContext>).mockReturnValue({
     playSound,
+    playSoundLooping: vitest.fn(),
+    soundDuration: {},
   });
 
   const { getByText, rtcSession } = createGroupCallView(

src/room/GroupCallView.tsx

@@ -53,7 +53,10 @@ import { InviteModal } from "./InviteModal";
 import { HeaderStyle, type UrlParams, useUrlParams } from "../UrlParams";
 import { E2eeType } from "../e2ee/e2eeType";
 import { useAudioContext } from "../useAudioContext";
-import { callEventAudioSounds } from "./CallEventAudioRenderer";
+import {
+  callEventAudioSounds,
+  type CallEventSounds,
+} from "./CallEventAudioRenderer";
 import { useLatest } from "../useLatest";
 import { usePageTitle } from "../usePageTitle";
 import {
@@ -317,8 +320,11 @@ export const GroupCallView: FC<Props> = ({
   const navigate = useNavigate();
 
   const onLeave = useCallback(
-    (cause: "user" | "error" = "user"): void => {
-      const audioPromise = leaveSoundContext.current?.playSound("left");
+    (
+      cause: "user" | "error" = "user",
+      playSound: CallEventSounds = "left",
+    ): void => {
+      const audioPromise = leaveSoundContext.current?.playSound(playSound);
       // In embedded/widget mode the iFrame will be killed right after the call ended prohibiting the posthog event from getting sent,
       // therefore we want the event to be sent instantly without getting queued/batched.
       const sendInstantly = !!widget;

src/room/InCallView.tsx

@@ -95,7 +95,10 @@ import {
 } from "../reactions/useReactionsSender";
 import { ReactionsAudioRenderer } from "./ReactionAudioRenderer";
 import { ReactionsOverlay } from "./ReactionsOverlay";
-import { CallEventAudioRenderer } from "./CallEventAudioRenderer";
+import {
+  CallEventAudioRenderer,
+  type CallEventSounds,
+} from "./CallEventAudioRenderer";
 import {
   debugTileLayout as debugTileLayoutSetting,
   useExperimentalToDeviceTransport as useExperimentalToDeviceTransportSetting,
@@ -117,11 +120,8 @@ import { Avatar, Size as AvatarSize } from "../Avatar";
 import waitingStyles from "./WaitingForJoin.module.css";
 import { prefetchSounds } from "../soundUtils";
 import { useAudioContext } from "../useAudioContext";
-// TODO: Dont use this!!!  use the correct sound
-import genericSoundOgg from "../sound/reactions/generic.ogg?url";
-import genericSoundMp3 from "../sound/reactions/generic.mp3?url";
-import leftCallSoundMp3 from "../sound/left_call.mp3";
-import leftCallSoundOgg from "../sound/left_call.ogg";
+import ringtoneMp3 from "../sound/ringtone.mp3?url";
+import ringtoneOgg from "../sound/ringtone.ogg?url";
 
 const canScreenshare = "getDisplayMedia" in (navigator.mediaDevices ?? {});
 
@@ -232,7 +232,7 @@ export interface InCallViewProps {
   livekitRoom: LivekitRoom;
   muteStates: MuteStates;
   /** Function to call when the user explicitly ends the call */
-  onLeave: () => void;
+  onLeave: (cause: "user", soundFile?: CallEventSounds) => void;
   header: HeaderStyle;
   otelGroupCallMembership?: OTelGroupCallMembership;
   connState: ECConnectionState;
@@ -281,14 +281,9 @@ export const InCallView: FC<InCallViewProps> = ({
   // Preload a waiting and decline sounds
   const pickupPhaseSoundCache = useInitial(async () => {
     return prefetchSounds({
-      waiting: { mp3: genericSoundMp3, ogg: genericSoundOgg },
-      decline: { mp3: leftCallSoundMp3, ogg: leftCallSoundOgg },
-      // Do we want a timeout sound?
+      waiting: { mp3: ringtoneMp3, ogg: ringtoneOgg },
     });
   });
-  // configure this to sth that fits to the pickup waiting sound.
-  // 1600 is in sync with the animation.
-  const PICKUP_SOUND_INTERVAL = 1600;
 
   const pickupPhaseAudio = useAudioContext({
     sounds: pickupPhaseSoundCache,
@@ -356,34 +351,47 @@ export const InCallView: FC<InCallViewProps> = ({
   const showFooter = useBehavior(vm.showFooter$);
   const earpieceMode = useBehavior(vm.earpieceMode$);
   const audioOutputSwitcher = useBehavior(vm.audioOutputSwitcher$);
-  useSubscription(vm.autoLeave$, onLeave);
+  useSubscription(vm.autoLeave$, () => onLeave("user"));
+
+  // We need to set the proper timings on the animation based upon the sound length.
+  const ringDuration = pickupPhaseAudio?.soundDuration["waiting"] ?? 1;
+  useEffect((): (() => void) => {
+    // The CSS animation includes the delay, so we must double the length of the sound.
+    window.document.body.style.setProperty(
+      "--call-ring-duration-s",
+      `${ringDuration * 2}s`,
+    );
+    window.document.body.style.setProperty(
+      "--call-ring-delay-s",
+      `${ringDuration}s`,
+    );
+    // Remove properties when we unload.
+    return () => {
+      window.document.body.style.removeProperty("--call-ring-duration-s");
+      window.document.body.style.removeProperty("--call-ring-delay-s");
+    };
+  }, [pickupPhaseAudio?.soundDuration, ringDuration]);
 
   // When we enter timeout or decline we will leave the call.
   useEffect((): void | (() => void) => {
     if (callPickupState === "timeout") {
-      onLeave();
+      onLeave("user", "timeout");
     }
     if (callPickupState === "decline") {
-      // Wait for the sound to finish before leaving
-      void pickupPhaseAudio
-        ?.playSound("decline")
-        .catch((e) => {
-          logger.error("Failed to play decline sound", e);
-        })
-        .finally(() => {
-          onLeave();
-        });
+      onLeave("user", "decline");
     }
   }, [callPickupState, onLeave, pickupPhaseAudio]);
 
   // When waiting for pickup, loop a waiting sound
   useEffect((): void | (() => void) => {
-    if (callPickupState !== "ringing") return;
-    const interval = window.setInterval(() => {
-      void pickupPhaseAudio?.playSound("waiting");
-    }, PICKUP_SOUND_INTERVAL);
-    return (): void => window.clearInterval(interval);
-  }, [callPickupState, pickupPhaseAudio]);
+    if (callPickupState !== "ringing" || !pickupPhaseAudio) return;
+    const endSound = pickupPhaseAudio.playSoundLooping("waiting", ringDuration);
+    return () => {
+      void endSound().catch((e) => {
+        logger.error("Failed to stop ringing sound", e);
+      });
+    };
+  }, [callPickupState, pickupPhaseAudio, ringDuration]);
 
   // Waiting UI overlay
   const waitingOverlay: JSX.Element | null = useMemo(() => {
@@ -823,7 +831,7 @@ export const InCallView: FC<InCallViewProps> = ({
     <EndCallButton
       key="end_call"
       onClick={function (): void {
-        onLeave();
+        onLeave("user");
       }}
       onTouchEnd={onControlsTouchEnd}
       data-testid="incall_leave"

src/room/ReactionAudioRenderer.test.tsx

@@ -69,6 +69,8 @@ beforeEach(() => {
   playSound = vitest.fn();
   (useAudioContext as MockedFunction<typeof useAudioContext>).mockReturnValue({
     playSound,
+    playSoundLooping: vitest.fn(),
+    soundDuration: {},
   });
 });
 

src/room/WaitingForJoin.module.css

@@ -25,7 +25,9 @@
   inset: -12px;
   border-radius: 9999px;
   border: 12px solid rgba(255, 255, 255, 0.6);
-  animation: pulse 1.6s ease-out infinite;
+  animation: pulse var(--call-ring-duration-s) ease-out infinite;
+  animation-delay: 1s;
+  opacity: 0;
 }
 
 .text {
@@ -36,13 +38,24 @@
   0% {
     transform: scale(0.95);
     opacity: 0.7;
+    transform: scale(0);
+    opacity: 1;
   }
-  70% {
+  35% {
     transform: scale(1.15);
     opacity: 0.15;
   }
-  100% {
+  50% {
     transform: scale(1.2);
     opacity: 0;
   }
+  50.01% {
+    transform: scale(0);
+  }
+  85% {
+    transform: scale(0);
+  }
+  100% {
+    transform: scale(0);
+  }
 }

src/state/CallViewModel.ts

@@ -49,6 +49,7 @@ import {
   race,
   scan,
   skip,
+  skipWhile,
   startWith,
   switchAll,
   switchMap,
@@ -853,17 +854,6 @@ export class CallViewModel extends ViewModel {
     throttleTime(THROTTLE_SOUND_EFFECT_MS),
   );
 
-  public readonly leaveSoundEffect$ = this.userMedia$.pipe(
-    pairwise(),
-    filter(
-      ([prev, current]) =>
-        current.length <= MAX_PARTICIPANT_COUNT_FOR_SOUND &&
-        current.length < prev.length,
-    ),
-    map(() => {}),
-    throttleTime(THROTTLE_SOUND_EFFECT_MS),
-  );
-
   /**
    * The number of participants currently in the call.
    *
@@ -963,7 +953,9 @@ export class CallViewModel extends ViewModel {
    *  - "success": Someone else joined. The call is in a normal state. No audiovisual feedback.
    *  - null: EC is configured to never show any waiting for answer state.
    */
-  public readonly callPickupState$ = this.options.waitForCallPickup
+  public readonly callPickupState$: Behavior<
+    "unknown" | "ringing" | "timeout" | "decline" | "success" | null
+  > = this.options.waitForCallPickup
     ? this.scope.behavior<
         "unknown" | "ringing" | "timeout" | "decline" | "success"
       >(
@@ -983,6 +975,24 @@ export class CallViewModel extends ViewModel {
       )
     : constant(null);
 
+  public readonly leaveSoundEffect$ = combineLatest([
+    this.callPickupState$,
+    this.userMedia$,
+  ]).pipe(
+    // Until the call is successful, do not play a leave sound.
+    // If callPickupState$ is null, then we always play the sound as it will not conflict with a decline sound.
+    skipWhile(([c]) => c !== null && c !== "success"),
+    map(([, userMedia]) => userMedia),
+    pairwise(),
+    filter(
+      ([prev, current]) =>
+        current.length <= MAX_PARTICIPANT_COUNT_FOR_SOUND &&
+        current.length < prev.length,
+    ),
+    map(() => {}),
+    throttleTime(THROTTLE_SOUND_EFFECT_MS),
+  );
+
   /**
    * List of MediaItems that we want to display, that are of type ScreenShare
    */

src/useAudioContext.tsx

@@ -32,6 +32,8 @@ async function playSound(
   buffer: AudioBuffer,
   volume: number,
   stereoPan: number,
+  delayS = 0,
+  abort?: AbortController,
 ): Promise<void> {
   const gain = ctx.createGain();
   gain.gain.setValueAtTime(volume, 0);
@@ -39,13 +41,62 @@ async function playSound(
   pan.pan.setValueAtTime(stereoPan, 0);
   const src = ctx.createBufferSource();
   src.buffer = buffer;
-  src.connect(gain).connect(pan).connect(ctx.destination);
+  abort?.signal.addEventListener("abort", () => {
+    src.disconnect();
+  });
   const p = new Promise<void>((r) => src.addEventListener("ended", () => r()));
+  src.connect(gain).connect(pan).connect(ctx.destination);
   controls.setPlaybackStarted();
-  src.start();
+  src.start(ctx.currentTime + delayS);
   return p;
 }
 
+/**
+ * Play a sound though a given AudioContext, looping until stopped. Will take
+ * care of connecting the correct buffer and gating
+ * through gain.
+ * @param volume The volume to play at.
+ * @param ctx The context to play through.
+ * @param buffer The buffer to play.
+ * @returns A function used to end the sound. This function will return a promise when the sound has stopped.
+ */
+function playSoundLooping(
+  ctx: AudioContext,
+  buffer: AudioBuffer,
+  volume: number,
+  stereoPan: number,
+  delayS?: number,
+): () => Promise<void> {
+  if (delayS === 0) {
+    throw Error("Looping sounds must have a delay");
+  }
+
+  // Our audio loop
+  let lastSoundPromise: Promise<void>;
+  let nextSoundPromise: Promise<void>;
+  let ac: AbortController | undefined;
+  void (async (): Promise<void> => {
+    ac = new AbortController();
+    // Play a sound immediately
+    lastSoundPromise = Promise.resolve();
+    do {
+      // Queue up the next sound.
+      nextSoundPromise = playSound(ctx, buffer, volume, stereoPan, delayS, ac);
+      // Await the previous sound.
+      await lastSoundPromise;
+      // Swap the promises over, and loop round to play the next sound.
+      lastSoundPromise = nextSoundPromise;
+    } while (!ac.signal.aborted);
+  })();
+
+  return async () => {
+    ac?.abort();
+    // Wait for sounds to finish.
+    await lastSoundPromise;
+    await nextSoundPromise;
+  };
+}
+
 interface Props<S extends string> {
   /**
    * The sounds to play. If no sounds should be played then
@@ -57,8 +108,13 @@ interface Props<S extends string> {
   muted?: boolean;
 }
 
-interface UseAudioContext<S> {
+interface UseAudioContext<S extends string> {
   playSound(soundName: S): Promise<void>;
+  playSoundLooping(soundName: S, delayS?: number): () => Promise<void>;
+  /**
+   * Map of sound name to duration in seconds.
+   */
+  soundDuration: Record<string, number>;
 }
 
 /**
@@ -146,5 +202,23 @@ export function useAudioContext<S extends string>(
         earpiecePan,
       );
     },
+    playSoundLooping: (name, delayS: number): (() => Promise<void>) => {
+      if (!audioBuffers[name]) {
+        throw Error(`Tried to play a sound that wasn't buffered (${name})`);
+      }
+      return playSoundLooping(
+        audioContext,
+        audioBuffers[name],
+        soundEffectVolume * earpieceVolume,
+        earpiecePan,
+        delayS,
+      );
+    },
+    soundDuration: Object.fromEntries(
+      Object.entries(audioBuffers).map(([k, v]) => [
+        k,
+        (v as AudioBuffer).duration,
+      ]),
+    ),
   };
 }

src/utils/test-viewmodel.ts

@@ -22,7 +22,10 @@ import {
 } from "matrix-js-sdk";
 
 import { E2eeType } from "../e2ee/e2eeType";
-import { CallViewModel } from "../state/CallViewModel";
+import {
+  CallViewModel,
+  type CallViewModelOptions,
+} from "../state/CallViewModel";
 import {
   mockLivekitRoom,
   mockMatrixRoom,
@@ -122,6 +125,7 @@ export function getBasicRTCSession(
 export function getBasicCallViewModelEnvironment(
   members: RoomMember[],
   initialRtcMemberships: CallMembership[] = [localRtcMember, aliceRtcMember],
+  callViewModelOptions: Partial<CallViewModelOptions> = {},
 ): {
   vm: CallViewModel;
   rtcMemberships$: BehaviorSubject<CallMembership[]>;
@@ -148,6 +152,7 @@ export function getBasicCallViewModelEnvironment(
     mockMediaDevices({}),
     {
       encryptionSystem: { kind: E2eeType.PER_PARTICIPANT },
+      ...callViewModelOptions,
     },
     of(ConnectionState.Connected),
     handRaisedSubject$,