It would be great if you can either provide the answer_audio instead of answer_snac or code to convert the answer_snac to an audio