Medix-AI/coqui-tts
Medix-AI
/
coqui-tts
mirror of https://github.com/coqui-ai/TTS.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

grave attention config update:

This commit is contained in:
root 2020-01-07 18:47:02 +01:00
parent 71af8da293
commit f2b6d00c45
3 changed files with 12 additions and 11 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

17
config.json
View File

@ -1,7 +1,7 @@
{ {
"model": "Tacotron2", // one of the model in models/ "model": "Tacotron2", // one of the model in models/
"run_name": "ljspeech", "run_name": "ljspeech-graves",
"run_description": "tacotron2 without bidirectional decoder", "run_description": "tacotron2 wuth graves attention",
// AUDIO PARAMETERS // AUDIO PARAMETERS
"audio":{ "audio":{
@ -38,7 +38,7 @@
"batch_size": 32, // Batch size for training. Lower values than 32 might cause hard to learn attention. It is overwritten by 'gradual_training'. "batch_size": 32, // Batch size for training. Lower values than 32 might cause hard to learn attention. It is overwritten by 'gradual_training'.
"eval_batch_size":16, "eval_batch_size":16,
"r": 7, // Number of decoder frames to predict per iteration. Set the initial values if gradual training is enabled. "r": 7, // Number of decoder frames to predict per iteration. Set the initial values if gradual training is enabled.
"gradual_training": [[0, 7, 64], [1, 5, 64], [50000, 3, 32], [130000, 2, 16], [290000, 1, 8]], // ONLY TACOTRON - set gradual training steps [first_step, r, batch_size]. If it is null, gradual training is disabled. "gradual_training": [[0, 7, 64], [1, 5, 64], [50000, 3, 32], [130000, 2, 32], [290000, 1, 32]], // ONLY TACOTRON - set gradual training steps [first_step, r, batch_size]. If it is null, gradual training is disabled.
"loss_masking": true, // enable / disable loss masking against the sequence padding. "loss_masking": true, // enable / disable loss masking against the sequence padding.
// VALIDATION // VALIDATION
@ -47,6 +47,7 @@
"test_sentences_file": null, // set a file to load sentences to be used for testing. If it is null then we use default english sentences. "test_sentences_file": null, // set a file to load sentences to be used for testing. If it is null then we use default english sentences.
// OPTIMIZER // OPTIMIZER
"noam_schedule": false,
"grad_clip": 1, // upper limit for gradients for clipping. "grad_clip": 1, // upper limit for gradients for clipping.
"epochs": 1000, // total number of epochs to train. "epochs": 1000, // total number of epochs to train.
"lr": 0.0001, // Initial learning rate. If Noam decay is active, maximum learning rate. "lr": 0.0001, // Initial learning rate. If Noam decay is active, maximum learning rate.
@ -60,7 +61,7 @@
"prenet_dropout": true, // enable/disable dropout at prenet. "prenet_dropout": true, // enable/disable dropout at prenet.
// ATTENTION // ATTENTION
"attention_type": "original", // 'original' or 'graves' "attention_type": "graves", // 'original' or 'graves'
"attention_heads": 5, // number of attention heads (only for 'graves') "attention_heads": 5, // number of attention heads (only for 'graves')
"attention_norm": "sigmoid", // softmax or sigmoid. Suggested to use softmax for Tacotron2 and sigmoid for Tacotron. "attention_norm": "sigmoid", // softmax or sigmoid. Suggested to use softmax for Tacotron2 and sigmoid for Tacotron.
"windowing": false, // Enables attention windowing. Used only in eval mode. "windowing": false, // Enables attention windowing. Used only in eval mode.
@ -90,8 +91,8 @@
"max_seq_len": 150, // DATASET-RELATED: maximum text length "max_seq_len": 150, // DATASET-RELATED: maximum text length
// PATHS // PATHS
// "output_path": "../keep/", // DATASET-RELATED: output path for all training outputs. "output_path": "/data5/rw/pit/keep/", // DATASET-RELATED: output path for all training outputs.
"output_path": "/media/erogol/data_ssd/Models/runs/", // "output_path": "/media/erogol/data_ssd/Models/runs/",
// PHONEMES // PHONEMES
"phoneme_cache_path": "mozilla_us_phonemes", // phoneme computation is slow, therefore, it caches results in the given folder. "phoneme_cache_path": "mozilla_us_phonemes", // phoneme computation is slow, therefore, it caches results in the given folder.
@ -108,8 +109,8 @@
[ [
{ {
"name": "ljspeech", "name": "ljspeech",
// "path": "/data/ro/shared/data/keithito/LJSpeech-1.1/", "path": "/data5/ro/shared/data/keithito/LJSpeech-1.1/",
"path": "/home/erogol/Data/LJSpeech-1.1", // "path": "/home/erogol/Data/LJSpeech-1.1",
"meta_file_train": "metadata_train.csv", "meta_file_train": "metadata_train.csv",
"meta_file_val": "metadata_val.csv" "meta_file_val": "metadata_val.csv"
} }

2
layers/common_layers.py
View File

@ -159,7 +159,7 @@ class GravesAttention(nn.Module):
k_t = gbk_t[:, 2, :] k_t = gbk_t[:, 2, :]
# attention GMM parameters # attention GMM parameters
sig_t = torch.nn.functional.softplus(b_t)+self.eps sig_t = torch.nn.functional.softplus(b_t) + self.eps
mu_t = self.mu_prev + torch.nn.functional.softplus(k_t) mu_t = self.mu_prev + torch.nn.functional.softplus(k_t)
g_t = torch.softmax(g_t, dim=-1) / sig_t + self.eps g_t = torch.softmax(g_t, dim=-1) / sig_t + self.eps

4
notebooks/ExtractTTSpectrogram.ipynb
View File

@ -303,9 +303,9 @@
"name": "python", "name": "python",
"nbconvert_exporter": "python", "nbconvert_exporter": "python",
"pygments_lexer": "ipython3", "pygments_lexer": "ipython3",
"version": "3.7.2" "version": "3.7.4"
} }
}, },
"nbformat": 4, "nbformat": 4,
"nbformat_minor": 2 "nbformat_minor": 4
} }