From 8631e286bd6ea5cf9af7d1d8fa3b8b4bac9fd5ee Mon Sep 17 00:00:00 2001
From: Dennis Thiessen <dennis@thiessen.io>
Date: Wed, 18 Feb 2026 11:52:41 +0100
Subject: [PATCH] added more logging for debugging voice issues

---
 backend/app/routers/voice.py          | 49 ++++++++++-----------------
 backend/app/services/voice_service.py | 11 ++++++
 frontend/js/voice.js                  |  5 +--
 3 files changed, 32 insertions(+), 33 deletions(-)

diff --git a/backend/app/routers/voice.py b/backend/app/routers/voice.py
index b21dd95..f6c7ae4 100644
--- a/backend/app/routers/voice.py
+++ b/backend/app/routers/voice.py
@@ -1,34 +1,13 @@
 """FluentGerman.ai — Admin voice-to-instruction & voice API router."""
 
-from fastapi import APIRouter, Depends, UploadFile, File
-from sqlalchemy.ext.asyncio import AsyncSession
+from fastapi import APIRouter, Depends, UploadFile, File, HTTPException
+import logging
 
-from app.auth import require_admin, get_current_user
-from app.config import get_settings
-from app.database import get_db
-from app.models import User
-from app.schemas import VoiceConfigOut, VoiceInstructionRequest
-from app.services.llm_service import summarize_instruction
-from app.services.voice_service import synthesize, transcribe
-from fastapi.responses import Response
+# ... imports ...
 
-router = APIRouter(prefix="/api/voice", tags=["voice"])
-
-
-@router.get("/config", response_model=VoiceConfigOut)
-async def voice_config(user: User = Depends(get_current_user)):
-    """Return current voice mode so frontend knows whether to use browser or API."""
-    settings = get_settings()
-    # API STT (Whisper) works with OpenAI-compatible providers
-    # Check if we have a dedicated voice key OR a generic LLM key for OpenAI
-    has_key = bool(settings.openai_api_key or (settings.llm_api_key and settings.llm_provider == "openai"))
-    
-    api_available = bool(settings.voice_mode == "api" and has_key)
-    return VoiceConfigOut(
-        voice_mode=settings.voice_mode,
-        voice_api_available=api_available,
-    )
+logger = logging.getLogger("fluentgerman.voice")
 
+# ...
 
 @router.post("/transcribe")
 async def transcribe_audio(
@@ -36,9 +15,13 @@ async def transcribe_audio(
     user: User = Depends(get_current_user),
 ):
     """Transcribe uploaded audio to text (API mode only)."""
-    audio_bytes = await audio.read()
-    text = await transcribe(audio_bytes, filename=audio.filename or "audio.webm")
-    return {"text": text}
+    try:
+        audio_bytes = await audio.read()
+        text = await transcribe(audio_bytes, filename=audio.filename or "audio.webm")
+        return {"text": text}
+    except Exception as e:
+        logger.error(f"Transcription failed: {str(e)}", exc_info=True)
+        raise HTTPException(status_code=500, detail=f"Transcription failed: {str(e)}")
 
 
 @router.post("/synthesize")
@@ -47,8 +30,12 @@ async def synthesize_text(
     user: User = Depends(get_current_user),
 ):
     """Convert text to speech audio (API mode only)."""
-    audio_bytes = await synthesize(text)
-    return Response(content=audio_bytes, media_type="audio/mpeg")
+    try:
+        audio_bytes = await synthesize(text)
+        return Response(content=audio_bytes, media_type="audio/mpeg")
+    except Exception as e:
+        logger.error(f"Synthesis failed: {str(e)}", exc_info=True)
+        raise HTTPException(status_code=500, detail=f"Synthesis failed: {str(e)}")
 
 
 @router.post("/generate-instruction", dependencies=[Depends(require_admin)])
diff --git a/backend/app/services/voice_service.py b/backend/app/services/voice_service.py
index c000cca..55aace1 100644
--- a/backend/app/services/voice_service.py
+++ b/backend/app/services/voice_service.py
@@ -1,6 +1,7 @@
 """FluentGerman.ai — Voice service (API provider + browser fallback)."""
 
 import io
+import logging
 
 import openai
 
@@ -12,6 +13,11 @@ async def transcribe(audio_bytes: bytes, filename: str = "audio.webm") -> str:
     settings = get_settings()
     # Use dedicated OpenAI key if available, otherwise fallback to LLM key
     api_key = settings.openai_api_key or settings.llm_api_key
+
+    key_type = "OPENAI_API_KEY" if settings.openai_api_key else "LLM_API_KEY"
+    masked = f"{api_key[:4]}...{api_key[-4:]}" if api_key and len(api_key) > 8 else "EMPTY"
+    logging.getLogger("fluentgerman.voice").info(f"Transcribing with {key_type}: {masked}")
+
     client = openai.AsyncOpenAI(api_key=api_key)
 
     audio_file = io.BytesIO(audio_bytes)
@@ -29,6 +35,11 @@ async def synthesize(text: str) -> bytes:
     settings = get_settings()
     # Use dedicated OpenAI key if available, otherwise fallback to LLM key
     api_key = settings.openai_api_key or settings.llm_api_key
+
+    key_type = "OPENAI_API_KEY" if settings.openai_api_key else "LLM_API_KEY"
+    masked = f"{api_key[:4]}...{api_key[-4:]}" if api_key and len(api_key) > 8 else "EMPTY"
+    logging.getLogger("fluentgerman.voice").info(f"Synthesizing with {key_type}: {masked}")
+
     client = openai.AsyncOpenAI(api_key=api_key)
 
     response = await client.audio.speech.create(
diff --git a/frontend/js/voice.js b/frontend/js/voice.js
index 758c10d..ed294b0 100644
--- a/frontend/js/voice.js
+++ b/frontend/js/voice.js
@@ -168,10 +168,11 @@ class VoiceManager {
                 this.lastInputWasVoice = true;
                 if (this.onResult) this.onResult(data.text);
             } else {
-                showToast('Transcription failed.', 'error');
+                const err = await response.json().catch(() => ({}));
+                showToast(`Transcription failed: ${err.detail || 'Unknown error'}`, 'error');
             }
         } catch (e) {
-            showToast('Transcription error', 'error');
+            showToast('Transcription network error', 'error');
         } finally {
             this.isRecording = false;
             if (this.onStateChange) this.onStateChange(false);