fix(stt): revert to reliable REST gpt-4o-transcribe + MediaRecorder full-blob (Realtime WS not accessible on key)

- Backend: added transcribe_audio (gpt-4o-transcribe), switched audio handler to full blob -> REST -> LLM -> streaming TTS
- Frontend: MediaRecorder (webm/opus) full recording sent on stop (one blob per utterance)
- Removed dead WhisperStream callbacks and pending_transcript/lock
- This unblocks voice per AUDIT item 1 (Option B fallback). Deltas will come in later item.
- Also preps for deprecation fix (MediaRecorder is the good path).

frontend/src/hooks/useConversation.ts

@@ -201,7 +201,7 @@ export function useConversation() {
 
     try {
       setMicError(null);
-      const stream = await navigator.mediaDevices.getUserMedia({ audio: true });
+      const stream = await navigator.mediaDevices.getUserMedia({ audio: { echoCancellation: true, noiseSuppression: true } });
       streamRef.current = stream;
 
       const ws = wsRef.current;
@@ -211,14 +211,27 @@ export function useConversation() {
         return;
       }
 
-      // PCM16 capture for Realtime WebSocket STT
-      captureRef.current = startPCMCapture(stream, (pcm16) => {
-        if (ws.readyState === WebSocket.OPEN) {
-          const base64 = arrayBufferToBase64(pcm16.buffer);
-          ws.send(JSON.stringify({ type: 'audio', data: base64 }));
+      // Use MediaRecorder for full utterance blob (Opus/webm) — sent on stop for REST STT
+      const mediaRecorder = new MediaRecorder(stream, { mimeType: 'audio/webm;codecs=opus' });
+      const chunks: Blob[] = [];
+      mediaRecorder.ondataavailable = (e) => {
+        if (e.data.size > 0) chunks.push(e.data);
+      };
+      mediaRecorder.onstop = () => {
+        if (chunks.length > 0 && ws.readyState === WebSocket.OPEN) {
+          const blob = new Blob(chunks, { type: 'audio/webm' });
+          blob.arrayBuffer().then((buf) => {
+            const base64 = arrayBufferToBase64(buf);
+            ws.send(JSON.stringify({ type: 'audio', data: base64 }));
+          });
         }
-      });
-
+        chunks.length = 0;
+        stream.getTracks().forEach((t) => t.stop());
+        streamRef.current = null;
+        setIsRecording(false);
+      };
+      recorderRef.current = mediaRecorder;
+      mediaRecorder.start();
       setIsRecording(true);
     } catch (err) {
       const msg = err instanceof Error ? err.message : String(err);
@@ -228,11 +241,16 @@ export function useConversation() {
   }, [addMessage]);
 
   const stopRecording = useCallback(() => {
-    captureRef.current?.stop();
-    captureRef.current = null;
-    streamRef.current?.getTracks().forEach((t) => t.stop());
-    streamRef.current = null;
-    setIsRecording(false);
+    if (recorderRef.current && recorderRef.current.state === 'recording') {
+      recorderRef.current.stop();
+      // onstop will handle sending the blob and cleanup
+    } else {
+      // fallback cleanup
+      streamRef.current?.getTracks().forEach((t) => t.stop());
+      streamRef.current = null;
+      setIsRecording(false);
+    }
+    captureRef.current = null; // legacy
   }, []);
 
   // ── Text ──
@@ -249,8 +267,8 @@ export function useConversation() {
     connect();
     return () => {
       wsRef.current?.close();
+      if (recorderRef.current && recorderRef.current.state === 'recording') recorderRef.current.stop();
       captureRef.current?.stop();
-      recorderRef.current?.stop();
       streamRef.current?.getTracks().forEach((t) => t.stop());
     };
   }, [connect]);