infer/lib/audio.py

import platform
import os
import ffmpeg
import numpy as np
import av
import traceback
from io import BytesIO

def wav2(i, o, format):
    inp = av.open(i, "rb")
    if format == "m4a":
        format = "mp4"
    out = av.open(o, "wb", format=format)
    if format == "ogg":
        format = "libvorbis"
    elif format == "mp4":
        format = "aac"

    ostream = out.add_stream(format)

    for frame in inp.decode(audio=0):
        for p in ostream.encode(frame):
            out.mux(p)

    for p in ostream.encode(None):
        out.mux(p)

    out.close()
    inp.close()

def load_audio(file, sr):
    try:
        # Clean the file path
        file = clean_path(file)
        if not os.path.exists(file):
            raise RuntimeError(
                "You input a wrong audio path that does not exist, please fix it!"
            )
        out, _ = (
            ffmpeg.input(file, threads=0)
            .output("-", format="f32le", acodec="pcm_f32le", ac=1, ar=sr)
            .run(cmd=["ffmpeg", "-nostdin"], capture_stdout=True, capture_stderr=True)
        )
    except Exception as e:
        traceback.print_exc()
        raise RuntimeError(f"Failed to load audio: {e}")

    audio_data = np.frombuffer(out, np.float32).flatten()
    
    # Ensure the function always returns a numpy array, even if it's empty
    if audio_data is None or len(audio_data) == 0:
        raise RuntimeError("Failed to load audio: no data returned.")
    
    return audio_data

def clean_path(path_str):
    if platform.system() == "Windows":
        path_str = path_str.replace("/", "\\")
    return path_str.strip().strip('"').strip("\n").strip('"').strip()

# Example usage
if __name__ == "__main__":
    try:
        audio_data = load_audio("example.wav", 16000)
        print("Audio data loaded successfully")
    except RuntimeError as e:
        print(e)