diff --git a/src/room/InCallView.tsx b/src/room/InCallView.tsx
index 4f83284b..c6d380eb 100644
--- a/src/room/InCallView.tsx
+++ b/src/room/InCallView.tsx
@@ -312,7 +312,7 @@ export const InCallView: FC<InCallViewProps> = ({
   const showHeader = useObservableEagerState(vm.showHeader$);
   const showFooter = useObservableEagerState(vm.showFooter$);
   const earpieceMode = useObservableEagerState(vm.earpieceMode$);
-  const toggleEarpieceMode = useObservableEagerState(vm.toggleEarpieceMode$);
+  const audioOutputSwitcher = useObservableEagerState(vm.audioOutputSwitcher$);
   const switchCamera = useSwitchCamera(vm.localVideo$);
 
   // Ideally we could detect taps by listening for click events and checking
@@ -457,27 +457,26 @@ export const InCallView: FC<InCallViewProps> = ({
 
   useAppBarSecondaryButton(
     useMemo(() => {
-      if (toggleEarpieceMode === null) return null;
-      const Icon = earpieceMode ? VolumeOnSolidIcon : EarpieceIcon;
+      if (audioOutputSwitcher === null) return null;
+      const isEarpieceTarget = audioOutputSwitcher.targetOutput === "earpiece";
+      const Icon = isEarpieceTarget ? EarpieceIcon : VolumeOnSolidIcon;
+      const label = isEarpieceTarget
+        ? t("settings.devices.earpiece")
+        : t("settings.devices.loudspeaker");
+
       return (
-        <Tooltip
-          label={
-            earpieceMode
-              ? t("settings.devices.earpiece")
-              : t("settings.devices.loudspeaker")
-          }
-        >
+        <Tooltip label={label}>
           <IconButton
             onClick={(e) => {
               e.preventDefault();
-              toggleEarpieceMode();
+              audioOutputSwitcher.switch();
             }}
           >
             <Icon />
           </IconButton>
         </Tooltip>
       );
-    }, [t, earpieceMode, toggleEarpieceMode]),
+    }, [t, audioOutputSwitcher]),
   );
 
   useAppBarHidden(!showHeader);
@@ -789,7 +788,7 @@ export const InCallView: FC<InCallViewProps> = ({
       <ReactionsAudioRenderer vm={vm} muted={muteAllAudio} />
       <EarpieceOverlay
         show={earpieceMode}
-        onBackToVideoPressed={toggleEarpieceMode}
+        onBackToVideoPressed={audioOutputSwitcher?.switch}
       />
       <ReactionsOverlay vm={vm} />
       {footer}
diff --git a/src/state/CallViewModel.ts b/src/state/CallViewModel.ts
index 9caa0925..fc1222c4 100644
--- a/src/state/CallViewModel.ts
+++ b/src/state/CallViewModel.ts
@@ -1261,29 +1261,36 @@ export class CallViewModel extends ViewModel {
 
   /**
    * Callback to toggle between the earpiece and the loudspeaker.
+   *
+   * This will be `null` in case the target does not exist in the list
+   * of available audio outputs.
    */
-  public readonly toggleEarpieceMode$: Observable<(() => void) | null> =
-    combineLatest(
-      [
-        this.mediaDevices.audioOutput.available$,
-        this.mediaDevices.audioOutput.selected$,
-      ],
-      (available, selected) => {
-        const selectionType = selected && available.get(selected.id)?.type;
-        if (!(selectionType === "speaker" || selectionType === "earpiece"))
-          return null;
+  public readonly audioOutputSwitcher$: Observable<{
+    targetOutput: "earpiece" | "speaker";
+    switch: () => void;
+  } | null> = combineLatest(
+    [
+      this.mediaDevices.audioOutput.available$,
+      this.mediaDevices.audioOutput.selected$,
+    ],
+    (available, selected) => {
+      const selectionType = selected && available.get(selected.id)?.type;
 
-        const newSelectionType =
-          selectionType === "speaker" ? "earpiece" : "speaker";
-        const newSelection = [...available].find(
-          ([, d]) => d.type === newSelectionType,
-        );
-        if (newSelection === undefined) return null;
+      // If we are in any output mode other than spaeker switch to speaker.
+      const newSelectionType =
+        selectionType === "speaker" ? "earpiece" : "speaker";
+      const newSelection = [...available].find(
+        ([, d]) => d.type === newSelectionType,
+      );
+      if (newSelection === undefined) return null;
 
-        const [id] = newSelection;
-        return () => this.mediaDevices.audioOutput.select(id);
-      },
-    );
+      const [id] = newSelection;
+      return {
+        targetOutput: newSelectionType,
+        switch: () => this.mediaDevices.audioOutput.select(id),
+      };
+    },
+  );
 
   public readonly reactions$ = this.reactionsSubject$.pipe(
     map((v) =>