-
It looks like in `train_speaker_embeddings.py` for the SpeakerRec recipe, the [following line] can cause this error:
```sh
Original Traceback (most recent call last):
File "/home/allabana/.virt…
-
Anyone can help with the train list and evaluation list?
From where I can find these both lists?
@joonson @dvisockas
-
**The bug**
When I tried to fine-tune pretrained models / train the SAD pipeline from scratch on my own data, the training process is always killed at the second epoch:
**To Reproduce**
```bash
$ …
-
The parent repository is using an MIT license. Can we assume that this repo is also licensed to MIT?
-
-
您好，
感谢您百忙之中查看我的邮件。
在kaldi中，训练神经网络之前会将特征生成用于训练网络的egs特征，这个值默认为[200,400]之间的随机数。
我的理解是这个将原始训练特征，切分为随机取值的过程，是为了增强网络对不同长度音频xvector提取的鲁棒性。
在您的asv-subtools中，好像是min-chunk直接设置的固定值。也就是用于训练的egs大小固定。
请问：
1.e…
-
Hello there,
Found some issues when playing with speaker verification recipes:
1. I found Line 241 in speaker_verification_cosine.py is trying to pass a 'rand_seed' argument to function prepar…
-
Can anyone confirm to have reproduced the results, **with the expected training speed (one epoch in around 5 to 10 minutes, using 1 GPU, on VoxCeleb-2 dataset)**?
Can you please tell me the exact ve…
-
This issue does not generate any error, but maybe confusion about the use:
https://github.com/speechbrain/speechbrain/blob/97e027ce5234e49e3f4969b218f0e12ee104f2cc/recipes/VoxCeleb/voxceleb_prepare…
-
Dear authors:
Thanks for your shared with such interesting work. In my experiment, I found that training the attention backend model with in-domain training data ( which means the data were used in t…