From b69da072d6e24955ed4bfd9e8a60a63a94f62721 Mon Sep 17 00:00:00 2001
From: sergiopesch <sergiopesch@users.noreply.github.com>
Date: Tue, 21 Apr 2026 12:32:07 +0200
Subject: [PATCH] fix(voice): reset finalization gate on new turn, attach
 remote audio to DOM

Two small but real issues in useVoiceInteraction.ts:

1. Dropped turn on fast re-speak.
   finalizeTurn() sets isFinalizingTurnRef.current = true, then awaits
   onFinalTranscript (a fetch to /api/chat). If the user starts a new
   turn before that fetch resolves, the server-emitted
   'input_audio_buffer.speech_started' event clears the buffers but
   does NOT clear the gate. The next 'input_audio_transcription.
   completed' event then hits:
       if (isFinalizingTurnRef.current) return;
   and gets silently dropped. The user's second utterance is lost.

   Fix: reset isFinalizingTurnRef to false on speech_started. The
   previous turn's in-flight fetch still completes and updates state;
   it just no longer blocks the next turn from finalizing.

2. Remote <audio> element not in the DOM.
   On pc.ontrack we lazily created an HTMLAudioElement with new Audio()
   and never appended it to the document. Most desktop browsers will
   play audio from a detached element, but Safari on iOS is known to
   be inconsistent -- autoplay can silently fail and audio-route
   selection (speaker vs earpiece vs Bluetooth) can misbehave on
   detached nodes.

   Fix: append to document.body with display:none + aria-hidden on
   first creation, and remove the element from its parent during
   teardown so repeated sessions don't leave orphaned <audio> nodes.

No changes to the public hook API. Verified with type-check, lint,
prettier, tests (12/12), production build.
---
 hooks/useVoiceInteraction.ts | 25 +++++++++++++++++++++++--
 1 file changed, 23 insertions(+), 2 deletions(-)

diff --git a/hooks/useVoiceInteraction.ts b/hooks/useVoiceInteraction.ts
index 0280df0..67f0e36 100644
--- a/hooks/useVoiceInteraction.ts
+++ b/hooks/useVoiceInteraction.ts
@@ -315,6 +315,14 @@ export function useVoiceInteraction({
       }
 
       if (event.type === 'input_audio_buffer.speech_started') {
+        // A fresh user turn is beginning. Reset the finalization gate so
+        // that if a previous turn's `onFinalTranscript` hasn't resolved yet
+        // (it's an awaited fetch), the upcoming
+        // `input_audio_transcription.completed` event isn't silently
+        // dropped by the `if (isFinalizingTurnRef.current) return;` guard.
+        // The previous turn's in-flight fetch will still complete and
+        // update state; it just won't block the next turn from finalizing.
+        isFinalizingTurnRef.current = false;
         finalBufferRef.current = '';
         assistantBufferRef.current = '';
         lastFinalizedTranscriptRef.current = '';
@@ -581,8 +589,18 @@ export function useVoiceInteraction({
       const remoteStream = event.streams[0];
       if (!remoteStream) return;
       if (!remoteAudioRef.current) {
-        remoteAudioRef.current = new Audio();
-        remoteAudioRef.current.autoplay = true;
+        const audio = new Audio();
+        audio.autoplay = true;
+        // Keep the element in the DOM. Some browsers (notably Safari on iOS)
+        // treat detached <audio> elements inconsistently for autoplay and
+        // audio-route selection -- rendering it (but visually hidden)
+        // avoids those edge cases without affecting layout.
+        audio.setAttribute('aria-hidden', 'true');
+        audio.style.display = 'none';
+        if (typeof document !== 'undefined' && document.body) {
+          document.body.appendChild(audio);
+        }
+        remoteAudioRef.current = audio;
       }
       remoteAudioRef.current.srcObject = remoteStream;
       remoteAudioRef.current.muted = isMuted;
@@ -785,6 +803,9 @@ export function useVoiceInteraction({
       if (remoteAudioRef.current) {
         remoteAudioRef.current.pause();
         remoteAudioRef.current.srcObject = null;
+        // Remove the element we appended to document.body on ontrack so
+        // repeated sessions don't leave orphaned <audio> nodes behind.
+        remoteAudioRef.current.parentNode?.removeChild(remoteAudioRef.current);
         remoteAudioRef.current = null;
       }
       micAnalyserRef.current = null;