Spaces:
Runtime error
Runtime error
| base_config: | |
| - egs/egs_bases/tts/vocoder/pwg.yaml | |
| - egs/egs_bases/tts/base_mel2wav.yaml | |
| - egs/datasets/audio/lj/pwg.yaml | |
| raw_data_dir: 'data/raw/LJSpeech-1.1' | |
| processed_data_dir: 'data/processed/LJSpeech_FastDiff' | |
| #binary_data_dir: 'data/binary/LJSpeech_Taco' | |
| binary_data_dir: /apdcephfs/private_nlphuang/preprocess/AdaGrad/data/binary/LJSpeech_Taco | |
| binarizer_cls: data_gen.tts.vocoder_binarizer.VocoderBinarizer | |
| pre_align_cls: egs.datasets.audio.lj.pre_align.LJPreAlign | |
| task_cls: modules.FastDiff.task.FastDiff.FastDiffTask | |
| binarization_args: | |
| with_wav: true | |
| with_spk_embed: false | |
| with_align: false | |
| with_word: false | |
| with_txt: false | |
| with_f0: false | |
| # data | |
| num_spk: 400 | |
| max_samples: 25600 | |
| aux_context_window: 0 | |
| max_sentences: 20 | |
| test_input_dir: '' # 'wavs' # wav->wav inference | |
| test_mel_dir: '' # 'mels' # mel->wav inference | |
| use_wav: True # mel->wav inference | |
| # training | |
| num_sanity_val_steps: -1 | |
| max_updates: 1000000 | |
| lr: 2e-4 | |
| weight_decay: 0 | |
| # FastDiff | |
| audio_channels: 1 | |
| inner_channels: 32 | |
| cond_channels: 80 | |
| upsample_ratios: [8, 8, 4] | |
| lvc_layers_each_block: 4 | |
| lvc_kernel_size: 3 | |
| kpnet_hidden_channels: 64 | |
| kpnet_conv_size: 3 | |
| dropout: 0.0 | |
| diffusion_step_embed_dim_in: 128 | |
| diffusion_step_embed_dim_mid: 512 | |
| diffusion_step_embed_dim_out: 512 | |
| use_weight_norm: True | |
| # Diffusion | |
| T: 1000 | |
| beta_0: 0.000001 | |
| beta_T: 0.01 | |
| noise_schedule: '' | |
| N: '' | |