search

resemble-ai / MelNet

WIP: Open Source Implementation of "MelNet: A Generative Model for Audio in the Frequency Domain"
MIT License
249 stars 50 forks source link

MelNet in TTS ? #1

Open kobenaxie opened 5 years ago

kobenaxie commented 5 years ago

https://github.com/fatchord/MelNet/blob/0929af0bf2d93bdb4e3539a3791a47d839ead63f/model.py#L86

In TTS, does the MelNet block work as Decoder in the Encoder-Decoder framework ? And if so, how can we integrate Attention in MelNet based TTS task ?

fatchord commented 5 years ago

@kobenaxie My plan for now is to get the unconditional model working first with upsampling tiers. After that I will add the attention/tts functionality. It might take a while though as I probably only have about 5-8 hours per week to work on this repo.