auto-git:
[add] dist/assets/index-DifKT69O.js [change] backend/main.py [change] backend/schemas.py [change] backend/whisper_admin.py [change] dist/index.html [change] electron/main.cjs [change] src/App.jsx [change] src/GeneralSettings.jsx [change] src/audioInput.js [unlink] dist/assets/index-Dm7DZNSo.js
This commit is contained in:
@@ -159,6 +159,7 @@ async def transcribe_audio_route(req: schemas.AudioTranscriptionRequest):
|
||||
audio_bytes,
|
||||
mime_type,
|
||||
req.model or DEFAULT_WHISPER_MODEL,
|
||||
req.language,
|
||||
)
|
||||
except RuntimeError as exc:
|
||||
raise HTTPException(status_code=400, detail=str(exc)) from exc
|
||||
|
||||
@@ -84,6 +84,7 @@ class AudioTranscriptionRequest(BaseModel):
|
||||
mime_type: str
|
||||
audio_base64: str
|
||||
model: Optional[str] = None
|
||||
language: Optional[str] = None
|
||||
|
||||
|
||||
class AudioTranscriptionResponse(BaseModel):
|
||||
|
||||
@@ -241,6 +241,7 @@ def transcribe_audio_bytes(
|
||||
audio_bytes: bytes,
|
||||
mime_type: str,
|
||||
model_name: str = DEFAULT_WHISPER_MODEL,
|
||||
language: Optional[str] = None,
|
||||
) -> Dict[str, Any]:
|
||||
if not audio_bytes:
|
||||
raise RuntimeError("Recorded audio was empty.")
|
||||
@@ -256,7 +257,14 @@ def transcribe_audio_bytes(
|
||||
_convert_audio_to_wav(input_path, wav_path)
|
||||
source_path = wav_path
|
||||
|
||||
result = model.transcribe(str(source_path), task="transcribe", fp16=device == "cuda")
|
||||
transcription_options = {
|
||||
"task": "transcribe",
|
||||
"fp16": device == "cuda",
|
||||
}
|
||||
if language:
|
||||
transcription_options["language"] = str(language).strip().lower()
|
||||
|
||||
result = model.transcribe(str(source_path), **transcription_options)
|
||||
return {
|
||||
"model": model_name,
|
||||
"device": device,
|
||||
|
||||
Reference in New Issue
Block a user