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

Merge branch 'dev-bos' into dev-tacotron2

This commit is contained in:
Eren Golge 2019-03-07 11:45:42 +01:00
parent 112fe0dc4d 772c859dff
commit cf11b6c23c
7 changed files with 42 additions and 36 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
.compute
View File

@ -3,5 +3,5 @@ ls ${SHARED_DIR}/data/keithito
pip3 install https://download.pytorch.org/whl/cu100/torch-1.0.1.post2-cp36-cp36m-linux_x86_64.whl pip3 install https://download.pytorch.org/whl/cu100/torch-1.0.1.post2-cp36-cp36m-linux_x86_64.whl
yes | apt-get install espeak yes | apt-get install espeak
python3 setup.py develop python3 setup.py develop
# python3 distribute.py --config_path config_cluster.json --data_path ${SHARED_DIR}/data/keithito/LJSpeech-1.1/ --restore_path ${USER_DIR}/best_model.pth.tar python3 distribute.py --config_path config_cluster.json --data_path ${SHARED_DIR}/data/keithito/LJSpeech-1.1/ --restore_path ${USER_DIR}/best_model.pth.tar
python3 distribute.py --config_path config_cluster.json --data_path ${SHARED_DIR}/data/keithito/LJSpeech-1.1/ # python3 distribute.py --config_path config_cluster.json --data_path ${SHARED_DIR}/data/keithito/LJSpeech-1.1/

9
config.json
View File

@ -1,6 +1,6 @@
{ {
"run_name": "queue", "run_name": "bos",
"run_description": "Queue memory and change lower r incrementatlly", "run_description": "bos character added to get away with the first char miss",
"audio":{ "audio":{
// Audio processing parameters // Audio processing parameters
@ -29,7 +29,7 @@
"url": "tcp:\/\/localhost:54321" "url": "tcp:\/\/localhost:54321"
}, },
"model": "Tacotron", // one of the model in models/. For now "Tacotron" or "Tacotron2" are available models. "model": "Tacotron2", // one of the model in models/
"grad_clip": 0.02, // upper limit for gradients for clipping. "grad_clip": 0.02, // 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.
@ -37,12 +37,11 @@
"warmup_steps": 4000, // Noam decay steps to increase the learning rate from 0 to "lr" "warmup_steps": 4000, // Noam decay steps to increase the learning rate from 0 to "lr"
"windowing": false, // Enables attention windowing. Used only in eval mode. "windowing": false, // Enables attention windowing. Used only in eval mode.
"memory_size": 5, // TO BE IMPLEMENTED -- memory queue size used to queue network predictions to feed autoregressive connection. Useful if r < 5. "memory_size": 5, // TO BE IMPLEMENTED -- memory queue size used to queue network predictions to feed autoregressive connection. Useful if r < 5.
"batch_group_size": 3,
"batch_size": 16, // Batch size for training. Lower values than 32 might cause hard to learn attention. "batch_size": 16, // Batch size for training. Lower values than 32 might cause hard to learn attention.
"eval_batch_size":16, "eval_batch_size":16,
"r": 1, // Number of frames to predict for step. "r": 1, // Number of frames to predict for step.
"wd": 0.000005, // Weight decay weight. "wd": 0.000002, // Weight decay weight.
"checkpoint": true, // If true, it saves checkpoints per "save_step" "checkpoint": true, // If true, it saves checkpoints per "save_step"
"save_step": 1000, // Number of training steps expected to save traning stats and checkpoints. "save_step": 1000, // Number of training steps expected to save traning stats and checkpoints.
"print_step": 10, // Number of steps to log traning on console. "print_step": 10, // Number of steps to log traning on console.

52
config_cluster.json
View File

@ -1,9 +1,8 @@
{ {
"model_name": "tts-master", "run_name": "bos",
"model_description": "tts master cluster test", "run_description": "bos character added to get away with the first char miss",
"audio":{ "audio":{
"audio_processor": "audio", // to use dictate different audio processors, if available.
// Audio processing parameters // Audio processing parameters
"num_mels": 80, // size of the mel spec frame. "num_mels": 80, // size of the mel spec frame.
"num_freq": 1025, // number of stft frequency levels. Size of the linear spectogram frame. "num_freq": 1025, // number of stft frequency levels. Size of the linear spectogram frame.
@ -20,41 +19,48 @@
"symmetric_norm": false, // move normalization to range [-1, 1] "symmetric_norm": false, // move normalization to range [-1, 1]
"max_norm": 1, // scale normalization to range [-max_norm, max_norm] or [0, max_norm] "max_norm": 1, // scale normalization to range [-max_norm, max_norm] or [0, max_norm]
"clip_norm": true, // clip normalized values into the range. "clip_norm": true, // clip normalized values into the range.
"mel_fmin": null, // minimum freq level for mel-spec. ~50 for male and ~95 for female voices. Tune for dataset!! "mel_fmin": 0.0, // minimum freq level for mel-spec. ~50 for male and ~95 for female voices. Tune for dataset!!
"mel_fmax": null, // maximum freq level for mel-spec. Tune for dataset!! "mel_fmax": 8000.0, // maximum freq level for mel-spec. Tune for dataset!!
"do_trim_silence": true // enable trimming of slience of audio as you load it. LJspeech (false), TWEB (false), Nancy (true) "do_trim_silence": false // enable trimming of slience of audio as you load it. LJspeech (false), TWEB (false), Nancy (true)
}, },
"embedding_size": 256, // Character embedding vector length. You don't need to change it in general. "distributed":{
"text_cleaner": "phoneme_cleaners", "backend": "nccl",
"url": "tcp:\/\/localhost:54321"
},
"model": "Tacotron2", // one of the model in models/
"grad_clip": 0.05, // 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.
"lr_decay": false, // if true, Noam learning rate decaying is applied through training. "lr_decay": false, // if true, Noam learning rate decaying is applied through training.
"loss_weight": 0.0, // loss weight to emphasize lower frequencies. Lower frequencies are in general more important for speech signals.
"warmup_steps": 4000, // Noam decay steps to increase the learning rate from 0 to "lr" "warmup_steps": 4000, // Noam decay steps to increase the learning rate from 0 to "lr"
"windowing": false, // Enables attention windowing. Used only in eval mode. "windowing": false, // Enables attention windowing. Used only in eval mode.
"memory_size": 5, // memory queue size used to queue network predictions to feed autoregressive connection. Useful if r < 5. "memory_size": 5, // TO BE IMPLEMENTED -- memory queue size used to queue network predictions to feed autoregressive connection. Useful if r < 5.
"batch_size": 32, // Batch size for training. Lower values than 32 might cause hard to learn attention. "batch_size": 16, // Batch size for training. Lower values than 32 might cause hard to learn attention.
"eval_batch_size":32, "eval_batch_size":16,
"r": 5, // Number of frames to predict for step. "r": 1, // Number of frames to predict for step.
"wd": 0.00001, // Weight decay weight. "wd": 0.000002, // Weight decay weight.
"checkpoint": true, // If true, it saves checkpoints per "save_step" "checkpoint": true, // If true, it saves checkpoints per "save_step"
"save_step": 5000, // Number of training steps expected to save traning stats and checkpoints. "save_step": 1000, // Number of training steps expected to save traning stats and checkpoints.
"print_step": 50, // Number of steps to log traning on console. "print_step": 10, // Number of steps to log traning on console.
"tb_model_param_stats": false, // true, plots param stats per layer on tensorboard. Might be memory consuming, but good for debugging. "tb_model_param_stats": false, // true, plots param stats per layer on tensorboard. Might be memory consuming, but good for debugging.
"batch_group_size": 8, //Number of batches to shuffle after bucketing.
"run_eval": true, "run_eval": true,
"test_delay_epochs": 100, //Until attention is aligned, testing only wastes computation time.
"data_path": "/media/erogol/data_ssd/Data/LJSpeech-1.1", // DATASET-RELATED: can overwritten from command argument "data_path": "/media/erogol/data_ssd/Data/LJSpeech-1.1", // DATASET-RELATED: can overwritten from command argument
"meta_file_train": "prompts_train.data", // DATASET-RELATED: metafile for training dataloader. "meta_file_train": "metadata_train.csv", // DATASET-RELATED: metafile for training dataloader.
"meta_file_val": "prompts_val.data", // DATASET-RELATED: metafile for evaluation dataloader. "meta_file_val": "metadata_val.csv", // DATASET-RELATED: metafile for evaluation dataloader.
"dataset": "nancy", // DATASET-RELATED: one of TTS.dataset.preprocessors depending on your target dataset. Use "tts_cache" for pre-computed dataset by extract_features.py "dataset": "ljspeech", // DATASET-RELATED: one of TTS.dataset.preprocessors depending on your target dataset. Use "tts_cache" for pre-computed dataset by extract_features.py
"min_seq_len": 0, // DATASET-RELATED: minimum text length to use in training "min_seq_len": 0, // DATASET-RELATED: minimum text length to use in training
"max_seq_len": 300, // DATASET-RELATED: maximum text length "max_seq_len": 1000, // DATASET-RELATED: maximum text length
"output_path": "models/", // DATASET-RELATED: output path for all training outputs. "output_path": "../keep/", // DATASET-RELATED: output path for all training outputs.
"num_loader_workers": 8, // number of training data loader processes. Don't set it too big. 4-8 are good values. "num_loader_workers": 8, // number of training data loader processes. Don't set it too big. 4-8 are good values.
"num_val_loader_workers": 4, // number of evaluation data loader processes. "num_val_loader_workers": 4, // number of evaluation data loader processes.
"phoneme_cache_path": "phonemes_cache", // phoneme computation is slow, therefore, it caches results in the given folder. "phoneme_cache_path": "ljspeech_us_phonemes", // phoneme computation is slow, therefore, it caches results in the given folder.
"use_phonemes": true, // use phonemes instead of raw characters. It is suggested for better pronounciation. "use_phonemes": true, // use phonemes instead of raw characters. It is suggested for better pronounciation.
"phoneme_language": "en-us" // depending on your target language, pick one from https://github.com/bootphon/phonemizer#languages "phoneme_language": "en-us", // depending on your target language, pick one from https://github.com/bootphon/phonemizer#languages
"text_cleaner": "phoneme_cleaners"
} }

2
distribute.py
View File

@ -131,7 +131,7 @@ def main(args):
Call train.py as a new process and pass command arguments Call train.py as a new process and pass command arguments
""" """
CONFIG = load_config(args.config_path) CONFIG = load_config(args.config_path)
OUT_PATH = create_experiment_folder(CONFIG.output_path, CONFIG.model_name, OUT_PATH = create_experiment_folder(CONFIG.output_path, CONFIG.run_name,
True) True)
stdout_path = os.path.join(OUT_PATH, "process_stdout/") stdout_path = os.path.join(OUT_PATH, "process_stdout/")

2
train.py
View File

@ -425,7 +425,7 @@ def main(args):
print( print(
" > Model restored from step %d" % checkpoint['step'], flush=True) " > Model restored from step %d" % checkpoint['step'], flush=True)
start_epoch = checkpoint['epoch'] start_epoch = checkpoint['epoch']
best_loss = checkpoint['postnet_loss'] # best_loss = checkpoint['postnet_loss']
args.restore_step = checkpoint['step'] args.restore_step = checkpoint['step']
else: else:
args.restore_step = 0 args.restore_step = 0

4
utils/text/__init__.py
View File

@ -43,7 +43,7 @@ def phoneme_to_sequence(text, cleaner_names, language):
''' '''
TODO: This ignores punctuations TODO: This ignores punctuations
''' '''
sequence = [] sequence = [_phonemes_to_id['^']]
clean_text = _clean_text(text, cleaner_names) clean_text = _clean_text(text, cleaner_names)
phonemes = text2phone(clean_text, language) phonemes = text2phone(clean_text, language)
# print(phonemes.replace('|', '')) # print(phonemes.replace('|', ''))
@ -81,7 +81,7 @@ def text_to_sequence(text, cleaner_names):
List of integers corresponding to the symbols in the text List of integers corresponding to the symbols in the text
''' '''
sequence = [] sequence = []
sequence = [_phonemes_to_id['^']]
# Check for curly braces and treat their contents as ARPAbet: # Check for curly braces and treat their contents as ARPAbet:
while len(text): while len(text):
m = _curly_re.match(text) m = _curly_re.match(text)

5
utils/text/symbols.py
View File

@ -9,6 +9,7 @@ from utils.text import cmudict
_pad = '_' _pad = '_'
_eos = '~' _eos = '~'
_bos = '^'
_characters = 'ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz!\'(),-.:;? ' _characters = 'ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz!\'(),-.:;? '
_punctuations = '!\'(),-.:;? ' _punctuations = '!\'(),-.:;? '
_phoneme_punctuations = '.!;:,?' _phoneme_punctuations = '.!;:,?'
@ -24,8 +25,8 @@ _phonemes = sorted(list(set(_phonemes)))
_arpabet = ['@' + s for s in _phonemes] _arpabet = ['@' + s for s in _phonemes]
# Export all symbols: # Export all symbols:
symbols = [_pad, _eos] + list(_characters) + _arpabet symbols = [_pad, _eos, _bos] + list(_characters) + _arpabet
phonemes = [_pad, _eos] + list(_phonemes) + list(_punctuations) phonemes = [_pad, _eos, _bos] + list(_phonemes) + list(_punctuations)
if __name__ == '__main__': if __name__ == '__main__':
print(" > TTS symbols ") print(" > TTS symbols ")