fix: accept uploaded wav voice notes

Co-authored-by: Codex <noreply@openai.com>

app.py

@@ -34,6 +34,7 @@ DATA_PATH = ROOT / "data" / "projects.json"
 INDEX_PATH = ROOT / "data" / "project_index.json"
 PROFILE_FIELDS = ["skills", "time", "preferences", "constraints"]
 MAX_AUDIO_UPLOAD_BYTES = 25 * 1024 * 1024
+AUDIO_UPLOAD_SUFFIXES = {".aac", ".aif", ".aiff", ".flac", ".m4a", ".mp3", ".oga", ".ogg", ".opus", ".wav", ".webm"}
 
 index = ProjectIndex.from_files(DATA_PATH, INDEX_PATH)
 engine = AdvisorEngine(index)
@@ -204,16 +205,24 @@ def agent_turn_stream(payload: dict[str, Any] | None = Body(default=None)) -> St
 @app.post("/api/transcribe")
 async def transcribe_audio(audio: UploadFile = File(...)) -> dict[str, Any]:
     content_type = str(audio.content_type or "")
-    if content_type and not content_type.startswith("audio/"):
+    filename = Path(str(audio.filename or "voice-note")).name
+    suffix = Path(filename).suffix.lower() or ".audio"
+    if not _is_audio_upload(content_type, suffix):
         raise HTTPException(status_code=415, detail="Voice input must be an audio file.")
     with tempfile.TemporaryDirectory(prefix="advisor-upload-") as directory:
-        filename = Path(str(audio.filename or "voice-note")).name
-        suffix = Path(filename).suffix or ".audio"
         source = Path(directory) / f"voice{suffix}"
         await _save_audio_upload(audio, source)
         return _transcribe_voice(str(source))
 
 
+def _is_audio_upload(content_type: str, suffix: str) -> bool:
+    if content_type.startswith("audio/"):
+        return True
+    if content_type in {"", "application/octet-stream"} and suffix in AUDIO_UPLOAD_SUFFIXES:
+        return True
+    return False
+
+
 async def _save_audio_upload(upload: UploadFile, target: Path) -> None:
     total = 0
     with target.open("wb") as handle:

tests/test_app.py

@@ -123,6 +123,26 @@ def test_transcribe_audio_endpoint_saves_audio(monkeypatch) -> None:
     assert captured["path"].endswith(".wav")
 
 
+def test_transcribe_audio_endpoint_accepts_octet_stream_audio(monkeypatch) -> None:
+    monkeypatch.setattr(
+        "app._transcribe_voice",
+        lambda path: {
+            "transcript": "A local-first memory archive.",
+            "model_id": "nvidia/nemotron-speech-streaming-en-0.6b",
+            "backend": "nemo-asr",
+            "sample_rate": 16000,
+        },
+    )
+
+    payload = asyncio.run(
+        transcribe_audio(
+            DummyUpload(b"RIFF....WAVE", filename="idea.wav", content_type="application/octet-stream")
+        )
+    )
+
+    assert payload["transcript"] == "A local-first memory archive."
+
+
 def test_transcribe_audio_endpoint_rejects_non_audio() -> None:
     upload = DummyUpload(b"hello", filename="note.txt", content_type="text/plain")