RuntimeError or FileFormatError

import whisper
import os

# This breaks when ffmpeg is missing or file format unsupported
audio_file = 'podcast.mp3'
model = whisper.load_model('base')
result = model.transcribe(audio_file)  # ❌ RuntimeError: Audio file format not supported
print(result['text'])

import whisper
import os
import subprocess
import json
from pathlib import Path

def transcribe_with_fallback(audio_path: str) -> str:
    """Transcribe audio with format validation and conversion fallback."""
    # Step 1: Verify ffmpeg is installed
    try:
        subprocess.run(['ffmpeg', '-version'], capture_output=True, check=True, timeout=5)
    except (FileNotFoundError, subprocess.TimeoutExpired):
        raise RuntimeError('ffmpeg not installed. Install with: brew install ffmpeg (macOS) or apt-get install ffmpeg (Linux)')
    
    # Step 2: Validate audio file format with ffprobe
    try:
        result = subprocess.run(
            ['ffprobe', '-v', 'error', '-show_format', '-show_streams', '-of', 'json', audio_path],
            capture_output=True,
            text=True,
            timeout=10
        )
        if result.returncode != 0:
            raise RuntimeError(f'ffprobe failed: {result.stderr}')
        probe_data = json.loads(result.stdout)
        if not probe_data.get('streams'):
            raise RuntimeError(f'No audio stream found in {audio_path} - file may be corrupted')
    except json.JSONDecodeError as e:
        raise RuntimeError(f'Failed to read audio file metadata: {e}')
    
    # Step 3: Convert to PCM s16 WAV if needed
    audio_path = Path(audio_path)
    wav_path = audio_path.with_suffix('.wav')
    
    if audio_path.suffix.lower() != '.wav':
        print(f'Converting {audio_path.suffix} to WAV...')
        subprocess.run(
            ['ffmpeg', '-i', str(audio_path), '-acodec', 'pcm_s16le', '-ac', '1', '-ar', '16000', '-y', str(wav_path)],
            capture_output=True,
            check=True
        )
        audio_path = wav_path
    
    # Step 4: Load and transcribe with error handling
    try:
        model = whisper.load_model('large-v3')  # Use latest model for best accuracy
        result = model.transcribe(str(audio_path))
        return result['text']
    except RuntimeError as e:
        raise RuntimeError(f'Whisper transcription failed on {audio_path}: {e}')

# Usage
try:
    text = transcribe_with_fallback('podcast.mp3')  # ✅ Now handles format conversion automatically
    print(f'Transcription: {text}')
except RuntimeError as e:
    print(f'Error: {e}')