Inquiry Regarding Audio Reconstruction Using vq_id

#2
by fish90510 - opened

Hello,

I'm very interested in your FAcodec and currently in the process of trying to encode and reconstruct audio files.
I've noticed that whether it's the v1 or v2 model, during the process of reconstructing audio, the code tends to utilize 'vq_post_emb' rather than 'vq_id'. Could you please provide guidance on how to solely use 'vq_id' for audio reconstruction?

Looking forward to your insights.

Best regards,
Fischer

Sign up or log in to comment