| audio: | |
| eps: 1e-10 | |
| fft_size: 2048 | |
| filter_length: 1200 | |
| hop_size: 300 | |
| log_base: 10.0 | |
| mel_fmax: 7600 | |
| mel_fmin: 80 | |
| num_mels: 80 | |
| sampling_rate: 24000 | |
| win_length: 1200 | |
| window: hann | |
| lang: | |
| - de | |
| - en | |
| model: | |
| decoder: | |
| conv_filter_size: 1024 | |
| conv_kernel_size: | |
| - 9 | |
| - 1 | |
| dropout: 0.2 | |
| n_head: 2 | |
| n_layers: 6 | |
| scln: true | |
| emb_dim: 512 | |
| emb_reduction: 1 | |
| encoder: | |
| depth: 2 | |
| expansion: 2 | |
| fs2_dropout: 0.2 | |
| fs2_head: 2 | |
| fs2_layer: 4 | |
| kernel_size: 5 | |
| kind: fastspeech2 | |
| n_heads: 2 | |
| ve_energy_quantization: linear | |
| ve_n_bins: 256 | |
| ve_pitch_quantization: linear | |
| vp_dropout: 0.5 | |
| vp_filter_size: 256 | |
| vp_kernel_size: 3 | |
| gst: | |
| n_heads: 8 | |
| n_style_tokens: 2000 | |
| ref_enc_filters: | |
| - 32 | |
| - 32 | |
| - 64 | |
| - 64 | |
| - 128 | |
| - 128 | |
| max_seq_len: 1500 | |
| postnet: | |
| postnet_embedding_dim: 0 | |
| postnet_kernel_size: 5 | |
| postnet_n_convolutions: 5 | |
| punct_emb_dim: 16 | |
| resnet: | |
| encoder_type: ASP | |
| layers: | |
| - 3 | |
| - 4 | |
| - 6 | |
| - 3 | |
| num_filters: | |
| - 32 | |
| - 64 | |
| - 128 | |
| - 256 | |
| spkemb: | |
| kind: ResNetSE34V2 | |
| stats: | |
| energy_max: 305.466064453125 | |
| energy_min: -2.440225667951865 | |
| pitch_max: 656.2979356469282 | |
| pitch_min: -45.333167047555264 | |