fix(talkbot): convert incoming audio to WAV before transcription

containers/talkbot/server.py

@@ -3,6 +3,7 @@ import logging
 import os
 import re
 import sqlite3
+import subprocess
 import threading
 from datetime import datetime, timedelta, timezone
 from typing import Optional
@@ -489,13 +490,59 @@ def is_blank_transcription(text: str) -> bool:
     }
 
 
+def convert_audio_to_wav(audio_bytes: bytes) -> bytes:
+    """Convert any FFmpeg-supported audio format to Whisper's WAV format."""
+    try:
+        result = subprocess.run(
+            [
+                "ffmpeg",
+                "-hide_banner",
+                "-loglevel",
+                "error",
+                "-i",
+                "pipe:0",
+                "-vn",
+                "-ac",
+                "1",
+                "-ar",
+                "16000",
+                "-c:a",
+                "pcm_s16le",
+                "-f",
+                "wav",
+                "pipe:1",
+            ],
+            input=audio_bytes,
+            capture_output=True,
+            check=False,
+            timeout=120,
+        )
+    except FileNotFoundError as exc:
+        raise RuntimeError("ffmpeg is not installed in the talkbot container") from exc
+    except subprocess.TimeoutExpired as exc:
+        raise RuntimeError("Audio conversion with ffmpeg timed out") from exc
+
+    if result.returncode != 0 or not result.stdout:
+        error = result.stderr.decode("utf-8", errors="replace").strip()
+        raise RuntimeError(f"Unable to convert audio to WAV: {error[:500]}")
+
+    log.info(
+        "Audio converted to WAV: %s -> %s bytes",
+        len(audio_bytes),
+        len(result.stdout),
+    )
+    return result.stdout
+
+
 def transcribe(audio_bytes: bytes) -> tuple[str, str]:
+    wav_bytes = convert_audio_to_wav(audio_bytes)
+
     response = httpx.post(
         f"{WHISPER_URL}/inference",
         files={
             "file": (
                 "audio.wav",
-                audio_bytes,
+                wav_bytes,
                 "audio/wav",
             )
         },
@@ -505,6 +552,12 @@ def transcribe(audio_bytes: bytes) -> tuple[str, str]:
         timeout=300,
     )
 
+    if response.is_error:
+        log.error(
+            "Whisper returned HTTP %s: %s",
+            response.status_code,
+            response.text[:500],
+        )
     response.raise_for_status()
 
     result = response.json()

containers/talkbot/talkbot.Containerfile

@@ -10,7 +10,7 @@ USER root
 EXPOSE 8100
 
 RUN apt-get update \
-    && apt-get install -y --no-install-recommends curl \
+    && apt-get install -y --no-install-recommends curl ffmpeg \
     && apt-get autoremove -y \
     && apt-get clean -y \
     && rm -rf /var/lib/apt/lists/*