bshall/UniversalVocoding

24kHz and 10 bit mu-law model

bshall opened this issue · 2 comments

Following the original paper train a model on 24kHz audio with 10 bit mu-law encoding.

thanks for your impressive work. Is the "24kHz and 10 bit mu-law model" training completed?

Hi @agangzz, unfortunately I had to move on to some other work so I haven't had time to look into training the 24kHz and 10 bit mu-law model. If you (or anyone else) are interested contributions would be more than welcome.

Note that the ZeroSpeech2019 data I used for the current model is 16kHz so it'd require a different multispeaker dataset.