pcm_to_mulaw.py

import numpy as np
import soundfile as sf
import matplotlib.pyplot as plt

def encode_mu_law(audio_data, mu=255):
    max_val = np.max(np.abs(audio_data))
    normalized_audio = audio_data / max_val
    encoded = np.sign(normalized_audio) * np.log(1 + mu * np.abs(normalized_audio)) / np.log(1 + mu)
    encoded_int16 = (encoded * 32767).astype(np.int16)
    
    print("Encoded stats: min={}, max={}, mean={}".format(
        np.min(encoded_int16), np.max(encoded_int16), np.mean(encoded_int16)))
    
    return encoded_int16, max_val

def decode_mu_law(encoded_audio, max_val, mu=255):
    normalized_encoded = encoded_audio.astype(np.float32) / 32767.0
    decoded = np.sign(normalized_encoded) * (1/mu) * ((1 + mu)**np.abs(normalized_encoded) - 1)
    decoded_audio = decoded * max_val
    return decoded_audio

def process_wav_file(input_path, output_encoded_path, output_decoded_path):
    audio_data, sample_rate = sf.read(input_path)
    
    print("Input audio stats: min={}, max={}, mean={}".format(
        np.min(audio_data), np.max(audio_data), np.mean(audio_data)))
    
    mu_law_encoded, max_val = encode_mu_law(audio_data)
    print(type(mu_law_encoded))
    sf.write(output_encoded_path, mu_law_encoded, sample_rate, subtype='PCM_16')
    np.save(output_encoded_path + '.maxval', max_val)
    
    decoded_audio = decode_mu_law(mu_law_encoded, max_val)
    sf.write(output_decoded_path, decoded_audio, sample_rate)
    
    print("Decoded audio stats: min={}, max={}, mean={}".format(
        np.min(decoded_audio), np.max(decoded_audio), np.mean(decoded_audio)))
    
    plt.figure(figsize=(12, 6))
    plt.subplot(2, 1, 1)
    plt.plot(audio_data)
    plt.title("Original Audio")
    plt.subplot(2, 1, 2)
    plt.plot(decoded_audio)
    plt.title("Decoded Audio (After μ-law round-trip)")
    plt.tight_layout()
    plt.show()


process_wav_file("input.wav", "output_mu_encoded.wav", "output_decoded.wav")