search

rishikksh20 / TalkNet2-pytorch

TalkNet 2: Non-Autoregressive Depth-Wise Separable Convolutional Model for Speech Synthesis with Explicit Pitch and Duration Prediction.
MIT License
85 stars 6 forks source link

TalkNet2 original code #1

Open Oktai15 opened 3 years ago

Oktai15 commented 3 years ago

Hello!

You can find original code for TalkNet2 here: https://github.com/NVIDIA/NeMo/blob/main/nemo/collections/tts/models/talknet.py Also take a look here: https://github.com/NVIDIA/NeMo/blob/main/tutorials/tts/3_TTS_TalkNet_Training.ipynb

Just for simplify re-implementation :)

rishikksh20 commented 3 years ago

@Oktai15 thanks !!!

rishikksh20 commented 3 years ago

ASR duration extractor and Gaussian Embedding seems very interesting.

m-toman commented 2 years ago

@Oktai15 cool, sorry to hijack this a bit but is the NeMo version updated to TalkNet2? Seems the code goes back a year or more where the TalkNet(1) paper was published. Interestingly the paper 2 does not mention the first one at all.

Pretty interested in it for embedded use. I got a pretty similar setup at the moment but more built around RNNs, pieces of Tacotron and a different prosody (pitch/duration/energy) model.

Oktai15 commented 2 years ago

@m-toman yes, NeMo code is updated to TalkNet2

mush42 commented 2 weeks ago

@Oktai15 seams the code no longer exists. Can you share it (if you still have it anyways).