-
作者你好:
我购买了9.9yuan模型并测试了算法,发现效果挺好的。但是我遇到2个女声样本,测试后发现非常难以区分:
```
----------- Configuration Arguments -----------
audio_path1: ../ysvoiceprint/voiceprint/zhou-tang/zhou3min-10-16k.wav
audio_pa…
-
-
I used DDP to run Common and Vox recipes with 4-8 Gpus, but instead of speeding up, it took longer to train one epoch.
My training environment is not NFS. Here is the script I ran
```
CUDA_VISIBLE_…
-
Hi!
Currently, I am extracting the embeddings from the ECAPA-TDNN with this code:
classifier = EncoderClassifier.from_hparams(source="speechbrain/spkrec-ecapa-voxceleb", run_opts={"device":"cuda"}…
-
hello:
I encountered a problem: how to train and call a model parameter of ecapa TDNN for speaker dialogue (AMI)
Please tell me some steps.
thank you !
-
Hi, author. I am training the ECAPA-TDNN model end-to-end by using: python trainECAPAModel.py
However, I found that while training, the training time per epoch is very long. After checking, I found …
-
I tried to relearn "lang id" with files with sample rate 8k. For that i resampled "CommonVoice" and "rirs_noise" datasets from 16k to 8k. In "hyperparams.yaml" i changed the sample rate to 8000. After…
-
Thank you for a good resource. :)
Is there any special reason to implement it differently from the original paper in the multi-scale (res2net) module part of the ECAPA-TDNN model?
(i.e., the first…
-
Hi, Firstly, thank u for this great project!
I want to export the pretrained ECAPA-TDNN model so that it can inference with C++ torch API.
When I tried to do "torch.jit.script(classifier.modules.emb…
-
Hello:
I have a problem in training speechtrain / recipes / voxceleb / speakerrec/_ speaker_ embeddings.py hparams/train_ ecapa_ I found that the configuration of each training is the same, but di…