Open Yocauu opened 6 months ago
It is found that individual methods in this set of neural networks, such as VQ-VAE, can use VQ-VAE2 to replace existing techniques
Through research, there are many possible methods in the vocoder section. We try to replace the original hifi-gan method with diffwave method which has better network structure to see if it can improve the quality of the generated audio.
The project currently does not have a mature separate integrated test document and needs to be improved in the future.
Observing that there is no mature data preprocessing in this project, which directly uses raw data for training, we propose to add data preprocessing