How to reconstruct audio directly from codes

[amuvarma13]

I use this function to get the codes.

with torch.no_grad():
    source_audio = librosa.load("/content/FAcodec/spks.wav", sr=24000)[0]
    source_audio = torch.tensor(source_audio).unsqueeze(0).float().to(device)
    z = model.encoder(source_audio[None, ...].to(device).float())
    z, quantized, _, _, timbre, codes = model.quantizer(z, source_audio[None, ...].to(device).float(), return_codes=True)

However, I am struggling to understand how to recover the audio from these codes and the timbre vector.