src/blu/asr.py

import os
import subprocess
from typing import Optional

from .config import cfg

def transcribe(audio_path: str, model: Optional[str] = None) -> str:
    """Transcribe using Whisper (via whisperx CLI for speed or openai-whisper).
    Requires: pip install whisperx OR openai-whisper
    """
    model = model or cfg.whisper_model

    try:
        cmd = ["whisperx", audio_path, "--model", model, "--output_format", "txt", "--output_dir", ".config"]
        subprocess.run(cmd, check=True, stdout=subprocess.PIPE, stderr=subprocess.PIPE)
        out = os.path.join(".config", os.path.splitext(os.path.basename(audio_path))[0] + ".txt")
        with open(out, "r") as f:
            return f.read().strip()
    except Exception:

        try:
            cmd = ["whisper", audio_path, "--model", model, "--language", "auto", "--output_format", "txt", "--output_dir", ".config"]
            subprocess.run(cmd, check=True, stdout=subprocess.PIPE, stderr=subprocess.PIPE)
            out = os.path.join(".config", os.path.splitext(os.path.basename(audio_path))[0] + ".txt")
            with open(out, "r") as f:
                return f.read().strip()
        except Exception as e:
            return f"ASR failed: {e}"