I’m posting this because I have a question.
If I git clone “clearer voice -studio” and run train.py in the train/speech_separation directory, I’d like to know whether the training starts from scratch using only the pre-trained framework, or if additional training is performed on the model from https://huggingface.co/alibabasglab/MossFormer2_SS_16K.
Also, if you happen to know, I’m curious about which dataset was used for the pre-training of the https://huggingface.co/alibabasglab/MossFormer2_SS_16K model.
I’m posting this because I have a question.
If I git clone “clearer voice -studio” and run
train.pyin thetrain/speech_separationdirectory, I’d like to know whether the training starts from scratch using only the pre-trained framework, or if additional training is performed on the model from https://huggingface.co/alibabasglab/MossFormer2_SS_16K.Also, if you happen to know, I’m curious about which dataset was used for the pre-training of the
https://huggingface.co/alibabasglab/MossFormer2_SS_16Kmodel.