Edresson Casanova comments

Results 41 comments of


                                            Edresson Casanova

[RFC][Feature request] Loading model onto specific GPU

Hi @jaketae, You can easily select the device using the "CUDA_VISIBLE_DEVICES=1". Using this only GPU 1 will be visible for the Pytorch and it will be forced to use this...

Exp 1. YourTTS-EN(VCTK) + SCL(speaker encoder layers are not initialized )

> @stalevna Hi! Did you manage to replicate the results of the first experiment? Recently, we created a recipe that replicates the first experiment proposed in the YourTTS paper. The...

From which version does coqui TTS starts supporting voice conversions and cloning?

> Thanks a lot! @Edresson I really appreciate the time you've taken out to answer this query. This will help me a lot, I just had one small query, that...

Fine tune YourTTS on own audio

@TejaswiniiB @etmyhome @mandray @leminhyen2 @e0xextazy @karynaur Recently, we created a recipe that makes everything easier. If you like you can try to fine-tune the model with this recipe. The recipe...

Fine Tune for Voice Conversion?

The training procedure for voice conversion and TTS are equal. If you like you can follow the recipe that replicates the first experiment proposed in the YourTTS paper. The recipe...

Languages other than PT, FR, EN

> @WeberJulian Hi! > > You mentioned that there will be a recipe to train a model on other languages using massively multilingual version of the model trained by you....

How to load WavLM ECAPA-TDNN embeddings for Speaker verification ?

@amitli1 Did you find any solution for this? I think the code is available here https://github.com/microsoft/UniSpeech/blob/e3043e2021d49429a406be09b9b8432febcdec73/downstreams/speaker_verification/models/ecapa_tdnn.py but I didn't find any checkpoint for it. A lot of papers are using...

[Bug] Wrong value for perceiver_cond_length_compression (256 instead of 1024)

Hi @talipturkmen, The conditioning input reducing factor when using the Perceiver Resampler is just 256 because we extract mel spectrogram with hop size equal to 256, check here: https://github.com/coqui-ai/TTS/blob/dbf1a08a0d4e47fdad6172e433eeb34bc6b13b4e/TTS/tts/layers/xtts/trainer/gpt_trainer.py#L147 Best...

Run test_run() every time that saves a checkpoint and not in the epochs end.

Hi @pranavpawar3 , contributions are always welcome on all Coqui repositories. Feel free to contribute :). @erogol This feature is not already implemented and is it welcome, right?

No Module Named 'torch'

I got the same issue here. I only was able to build from the source (clone the repo then run `python setup.py install`). `pip install git+https://github.com/HazyResearch/flash-attention` also give me the...