fix(elevenlabs_asr): __main__ block passes a language_code kwarg that does not exist

[Anai-Guo]

Problem

Running core/asr_backend/elevenlabs_asr.py directly fails immediately — its __main__
block calls the function with arguments that do not match its signature:

def transcribe_audio_elevenlabs(raw_audio_path, vocal_audio_path, start=None, end=None):  # L67
...
if __name__ == "__main__":
    file_path = input("Enter local audio file path (mp3 format): ")
    language = input("Enter language code for transcription (en or zh or other...): ")
    result = transcribe_audio_elevenlabs(file_path, language_code=language)   # L141

Two independent errors:

1. language_code is not a parameter →
   TypeError: transcribe_audio_elevenlabs() got an unexpected keyword argument 'language_code'
2. vocal_audio_path is required and never passed → missing a required argument

There is also a third, quieter problem: the script prompts the user for a language that
the function never consults. The request body sources it from config instead:

data = {
    "model_id": "scribe_v1",
    "timestamps_granularity": "word",
    "language_code": load_key("whisper.language"),   # <-- from config.yaml, not the argument

So the prompt collects a value that has no effect on the result.

The intended call shape

The real caller in core/_2_asr.py:41 passes all four positionally:

result = ts(_RAW_AUDIO_FILE, vocal_audio, start, end)

The __main__ block is the only call site that disagrees, which is why this has gone
unnoticed — the normal pipeline path works fine.

Fix

Mirror the real caller and drop the prompt that had no effect:

if __name__ == "__main__":
    file_path = input("Enter local audio file path (mp3 format): ")
    # Language is not a parameter: the request body reads whisper.language from config.
    result = transcribe_audio_elevenlabs(file_path, file_path)

Passing file_path for both raw_audio_path and vocal_audio_path matches the
standalone use case, where there is no separated vocal track — only vocal_audio_path
is actually loaded, and raw_audio_path is unused in the body.

Only the __main__ block changes; transcribe_audio_elevenlabs and the pipeline path
are untouched.

---

🤖 Generated with Claude Code