DiffWave Vocoder

This is a reimplementaion of the neural vocoder in DIFFWAVE: A VERSATILE DIFFUSION MODEL FOR AUDIO SYNTHESIS.

Usage:

• To continue training the model, run python distributed_train.py -c config_${channel}.json, where ${channel} can be either 64 or 128.

• To retrain the model, change the parameter ckpt_iter in the corresponding json file to -1 and use the above command.

• To generate audio, run python inference.py -c config_${channel}.json -cond ${conditioner_name}. For example, if the name of the mel spectrogram is LJ001-0001.wav.pt, then ${conditioner_name} is LJ001-0001. Provided mel spectrograms include LJ001-0001 through LJ001-0186.

• Note, you may need to carefully adjust some parameters in the json file, such as data_path and batch_size_per_gpu.

Pretrained models and generated samples:

• channel=64 model
• channel=64 samples
• channel=128 model
• channel=128 samples