Benjamin_Boenisch/breakpilot-compliance
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

feat(tts): add English voice (lessac-high) + language-based model selection

Browse Source 
- Download en_US-lessac-high Piper model in Dockerfile
- Select TTS engine based on request language (de/en)
- Include language in cache key to avoid collisions
- List both voices in /voices endpoint

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
This commit is contained in:
Benjamin Admin
2026-03-20 14:07:23 +01:00
parent 5ea31a3236
commit 4f6ac9b23a
2 changed files with 37 additions and 16 deletions
Download Patch File Download Diff File Expand all files Collapse all files

12
compliance-tts-service/Dockerfile
View File

@@ -15,8 +15,16 @@ WORKDIR /app
COPY requirements.txt .
RUN pip install --no-cache-dir -r requirements.txt
# Download Piper voice model (German, thorsten, high quality)
RUN mkdir -p /app/models && wget -q -O /app/models/de_DE-thorsten-high.onnx "https://huggingface.co/rhasspy/piper-voices/resolve/main/de/de_DE/thorsten/high/de_DE-thorsten-high.onnx" && wget -q -O /app/models/de_DE-thorsten-high.onnx.json "https://huggingface.co/rhasspy/piper-voices/resolve/main/de/de_DE/thorsten/high/de_DE-thorsten-high.onnx.json"
# Download Piper voice models
RUN mkdir -p /app/models && \
wget -q -O /app/models/de_DE-thorsten-high.onnx \
"https://huggingface.co/rhasspy/piper-voices/resolve/main/de/de_DE/thorsten/high/de_DE-thorsten-high.onnx" && \
wget -q -O /app/models/de_DE-thorsten-high.onnx.json \
"https://huggingface.co/rhasspy/piper-voices/resolve/main/de/de_DE/thorsten/high/de_DE-thorsten-high.onnx.json" && \
wget -q -O /app/models/en_US-lessac-high.onnx \
"https://huggingface.co/rhasspy/piper-voices/resolve/main/en/en_US/lessac/high/en_US-lessac-high.onnx" && \
wget -q -O /app/models/en_US-lessac-high.onnx.json \
"https://huggingface.co/rhasspy/piper-voices/resolve/main/en/en_US/lessac/high/en_US-lessac-high.onnx.json"
# Copy application
COPY . .

41
compliance-tts-service/main.py
View File

@@ -23,6 +23,7 @@ MINIO_ACCESS_KEY = os.getenv("MINIO_ACCESS_KEY", "breakpilot")
MINIO_SECRET_KEY = os.getenv("MINIO_SECRET_KEY", "breakpilot123")
MINIO_SECURE = os.getenv("MINIO_SECURE", "false").lower() == "true"
PIPER_MODEL_PATH = os.getenv("PIPER_MODEL_PATH", "/app/models/de_DE-thorsten-high.onnx")
PIPER_MODEL_EN_PATH = os.getenv("PIPER_MODEL_EN_PATH", "/app/models/en_US-lessac-high.onnx")
AUDIO_BUCKET = "compliance-training-audio"
VIDEO_BUCKET = "compliance-training-video"
@@ -30,6 +31,7 @@ VIDEO_BUCKET = "compliance-training-video"
# Initialize services
storage = StorageClient(MINIO_ENDPOINT, MINIO_ACCESS_KEY, MINIO_SECRET_KEY, secure=MINIO_SECURE)
tts = PiperTTS(PIPER_MODEL_PATH)
tts_en = PiperTTS(PIPER_MODEL_EN_PATH) if os.path.exists(PIPER_MODEL_EN_PATH) else None
@app.on_event("startup")
@@ -106,16 +108,22 @@ async def health():
@app.get("/voices")
async def list_voices():
"""List available TTS voices."""
return {
"voices": [
VoiceInfo(
id="de_DE-thorsten-high",
language="de",
name="Thorsten (High Quality)",
quality="high",
),
],
}
voices = [
VoiceInfo(
id="de_DE-thorsten-high",
language="de",
name="Thorsten (High Quality)",
quality="high",
),
]
if tts_en is not None:
voices.append(VoiceInfo(
id="en_US-lessac-high",
language="en",
name="Lessac (High Quality)",
quality="high",
))
return {"voices": voices}
class SynthesizeDirectRequest(BaseModel):
@@ -138,8 +146,8 @@ async def synthesize_direct(req: SynthesizeDirectRequest):
if not req.text.strip():
raise HTTPException(status_code=400, detail="Text is empty")
# Cache key based on text hash
text_hash = hashlib.sha256(req.text.encode()).hexdigest()[:16]
# Cache key based on text + language hash
text_hash = hashlib.sha256(f"{req.language}:{req.text}".encode()).hexdigest()[:16]
cache_path = os.path.join(TTS_CACHE_DIR, f"{text_hash}.mp3")
if os.path.exists(cache_path):
@@ -150,13 +158,18 @@ async def synthesize_direct(req: SynthesizeDirectRequest):
headers={"X-TTS-Cache": "hit"},
)
# Select TTS engine based on language
engine = tts
if req.language == "en" and tts_en is not None:
engine = tts_en
with tempfile.TemporaryDirectory() as tmpdir:
try:
mp3_path, duration = tts.synthesize_to_mp3(req.text, tmpdir)
mp3_path, duration = engine.synthesize_to_mp3(req.text, tmpdir)
# Copy to cache
import shutil
shutil.copy2(mp3_path, cache_path)
logger.info(f"TTS synthesized: {len(req.text)} chars, {duration:.1f}s, cached as {text_hash}")
logger.info(f"TTS synthesized ({req.language}): {len(req.text)} chars, {duration:.1f}s, cached as {text_hash}")
except Exception as e:
logger.error(f"Direct synthesis failed: {e}")
raise HTTPException(status_code=500, detail=str(e))