app.py

import os
import subprocess
import logging
from datetime import datetime

from denoiser import Denoiser
from transcriber import Transcriber
from translator import Translator

logger = logging.getLogger(__name__)

denoiser    = Denoiser()
transcriber = Transcriber()
translator  = Translator()

def run_pipeline(audio_input, remove_fillers, remove_stutters, remove_silences, remove_breaths, remove_mouth, target_lang):
    if not audio_input:
        return None, "No audio detected. Please upload a file.", ""

    # Create job directory
    job_id = datetime.now().strftime("%Y%m%d_%H%M%S")
    output_dir = f"outputs/{job_id}"
    os.makedirs(output_dir, exist_ok=True)
    
    try:
        # 1. DENOISE & CLEAN
        # The new 'process' method handles everything via the Cleanvoice API
        logger.info("[Pipeline] Step 1: Denoising & Studio Processing...")
        res = denoiser.process(
            audio_input, output_dir,
            fillers=remove_fillers, 
            stutters=remove_stutters,
            long_silences=remove_silences, 
            breath=remove_breaths,
            mouth_sounds=remove_mouth
        )
        
        # 2. TRANSCRIBE
        logger.info("[Pipeline] Step 2: Transcribing...")
        # FIX: transcribe() returns a tuple (text, lang, method), not a dict
        text, _lang, _method = transcriber.transcribe(res['audio_path'])
        
        # 3. TRANSLATE
        logger.info(f"[Pipeline] Step 3: Translating to {target_lang}...")
        translated = translator.translate(text, target_lang)

        # 4. FINAL EXPORT: Conversion to MP3
        final_mp3 = os.path.join(output_dir, "ClearWave_Studio_Final.mp3")
        subprocess.run([
            "ffmpeg", "-y", "-i", res['audio_path'],
            "-codec:a", "libmp3lame", "-qscale:a", "2", final_mp3
        ], check=True)

        return final_mp3, text, translated

    except Exception as e:
        logger.error(f"Pipeline Error: {e}")
        return None, f"Error: {str(e)}", ""