my_utils.py

import ffmpeg
import numpy as np
import requests
import logging
import wave
import librosa

def load_audio(file_path, sr):
    try:
        with wave.open(file_path, 'rb') as audio_file:
            channels = audio_file.getnchannels()
            sample_width = audio_file.getsampwidth()
            frame_rate = audio_file.getframerate()
            frames = audio_file.readframes(audio_file.getnframes())

        audio_data = np.frombuffer(frames, dtype=np.int16)
        audio_data = audio_data.astype(np.float32) / np.iinfo(np.int16).max  # Convert to float and normalize to the range [-1, 1]

        # Resample the audio if the sample rate is different
        if frame_rate != sr:
            audio_data = librosa.resample(audio_data, orig_sr=frame_rate, target_sr=sr)

        # Perform any required audio processing or conversion
        # ...

    except Exception as e:
        raise RuntimeError(f"Failed to load audio: {e}")

    return audio_data