auspicious3000 autovc issues

auspicious3000 / autovc

AutoVC: Zero-Shot Voice Style Transfer with Only Autoencoder Loss

https://arxiv.org/abs/1905.05879

MIT License

973 stars 206 forks source link

issues

Newest

Newest Most commented Recently updated Oldest Least commented Least recently updated

Unintelligible Results on Custom Data and on the sample /Wavs from this Repo

#124 jvel07 opened 2 months ago
0
关于损失函数的一些问题

#123 IndowK opened 9 months ago
5
Number of speakers used to train downloadable "pre-trained model"

#117 MelissaChen15 closed 1 year ago
2
An error on attempt to train model

#116 daaibraanies closed 1 year ago
1
An error is reported when the trained model is tested

#115 SeptemberN closed 1 year ago
3
How to save my own trained model？

#114 SeptemberN closed 2 years ago
1
How to generate metadata.pkl file ?

#113 Edward205 opened 2 years ago
6
About the one-hot ablation study

#112 Irislucent closed 2 years ago
1
Exchange librosa.output.write with soundfile.write (Librosa deprecated)

#111 lisabecker closed 2 years ago
1
How I can change Content wav?

#110 hjs5979 closed 2 years ago
1
How to use Pretrained model for Inference?

#109 ayush714 opened 2 years ago
2
How to test on my own data?

#108 Ha0Tang opened 2 years ago
7
len_crop issue when train with VCTK dataset

#107 junseokoh1 opened 2 years ago
2
hifi_gan sampling rate

#106 zjuPeco closed 2 years ago
1
Difference in calculating mel-spectrogram between AutoVC and vocoders

#105 Irislucent closed 2 years ago
1
why no validation?

#104 zjuPeco closed 2 years ago
2
After training, where is the trained autovc model

#103 leizetong closed 2 years ago
1
How to change speaker encoder to one-hot encoder

#102 Jwaminju opened 2 years ago
6
l_recon and l_recon0 never converges

#101 gkv856 opened 2 years ago
1
Differences in Architecture Between Code and Paper

#100 taubaaron opened 2 years ago
1
Pretrained model for speaker encoder

#99 anitaweng closed 2 years ago
5
downsample factor

#98 inconnu11 closed 2 years ago
1
differences in mel-spectogram

#97 amiteliav opened 2 years ago
4
How to use the new hifi-gan model

#96 antovespoli3 closed 2 years ago
4
Error when increasing batch_size

#95 vasyarv opened 2 years ago
2
fix shape mismatch in loss calculation.

#94 dodobyte opened 2 years ago
0
Test Set for evaluation in Paper

#93 v-nhandt21 opened 3 years ago
1
Consulting request

#92 dmks closed 3 years ago
0
Analysis window length (fft_length) and hop_length for feature extraction

#91 rppravin opened 3 years ago
1
Differences in Talker Embedding Extraction

#90 rppravin opened 3 years ago
0
F0 Converter for P - loss function values

#88 rishabhjain16 closed 3 years ago
1
Getting the voice conversion work on short segments (~120 ms)

#86 rppravin closed 3 years ago
1
Is there someone with knowledge who is willing to transfer two voices over another for me?

#85 Berendtol closed 2 years ago
0
How to get the same mel feature in "metadata.pkl"?

#84 gnipping opened 3 years ago
19
Where is the "Deconv Layer" in figure 3(d) in your paper?

#83 gnipping closed 3 years ago
4
How to generate test data

#82 ZengHorace closed 3 years ago
1
bugfix to make_metadata.py

#81 JohnHerry closed 3 years ago
0
How to get wavfrom from the predicted mels

#80 JohnHerry closed 3 years ago
0
I wrote a single Jupyter notebook to reproduce the results, but all I get is silence, please help

#79 ghost closed 3 years ago
7
Where is metadata.pkl created for my own dataset?

#78 ghost closed 3 years ago
2
Error(s) in loading state_dict for Generator: size mismatch for encoder.lstm.weight_ih_l0: copying a param with shape torch.Size([64, 512]) from checkpoint, the shape in current model is torch.Size([128, 512])

#77 ghost closed 3 years ago
6
The autovc.ckpt file size is 341MB but in training is 242MB

#76 ghost closed 3 years ago
0
In train.py where are the checkpoints saved?

#75 ghost closed 3 years ago
3
For training, how many speakers are required?

#73 ghost closed 3 years ago
1
problem with converting output

#72 todalex opened 3 years ago
0
How to reproduce the result on VCTK dataset?

#71 liangshuang1993 opened 3 years ago
1
Should dvector/autovc/wavenet-vocoder use the exactly same mel-spectrogram algorithm?

#70 KevinHua opened 3 years ago
1
Request info on training data used for pre-trained models

#69 rppravin opened 3 years ago
1
How to train in ONE-HOT pattern?

#68 Lukelluke opened 3 years ago
1
How to use this for repo for just testing?

#67 sandeshnaroju opened 3 years ago
11