modules/audio_processor.py

"""
Audio file validation and decoding.
"""
from __future__ import annotations

import logging
from pathlib import Path

import numpy as np

from config.settings import MAX_UPLOAD_SIZE_MB, SUPPORTED_EXTENSIONS

logger = logging.getLogger(__name__)

# Whisper and the diarization models both expect 16 kHz mono.
SAMPLE_RATE = 16000


# ---------------------------------------------------------------------------
# Validation
# ---------------------------------------------------------------------------
def validate_file_extension(filename: str) -> bool:
    """Check whether the file extension is in the supported set."""
    return Path(filename).suffix.lower() in SUPPORTED_EXTENSIONS


def validate_file_size(file_size_bytes: int) -> bool:
    """Return True if file is within the configured upload limit."""
    return file_size_bytes <= MAX_UPLOAD_SIZE_MB * 1024 * 1024


# ---------------------------------------------------------------------------
# Decoding
# ---------------------------------------------------------------------------
def load_audio(file_path: Path) -> np.ndarray:
    """
    Decode any supported audio/video file to 16 kHz mono float32 samples.

    Uses the FFmpeg libraries bundled with faster-whisper (PyAV), so no
    system FFmpeg install is needed. The result is decoded once and shared
    by transcription and diarization.
    """
    from faster_whisper import decode_audio

    try:
        audio = decode_audio(str(file_path), sampling_rate=SAMPLE_RATE)
    except Exception as exc:
        raise ValueError(
            "This file could not be read as audio. It may be corrupt or in an unsupported format."
        ) from exc
    if len(audio) == 0:
        raise ValueError("This file contains no audio.")
    logger.info("Decoded %.1f s of audio from %s", len(audio) / SAMPLE_RATE, file_path)
    return audio