Skip to content

Commit

Permalink
config updates, m-ai-labs processor update. If no metafileis given pa…
Browse files Browse the repository at this point in the history 
…rse the ffodler to use all the speakers
  • Loading branch information
Eren Golge committed Jun 21, 2019
1 parent b35b6e8 commit 118fe61
Show file tree
Hide file tree
Showing 3 changed files with 20 additions and 52 deletions.
2 changes: 1 addition & 1 deletion .compute
View file
Original file line number Diff line number Diff line change
Expand Up @@ -10,7 +10,7 @@ wget https://www.dropbox.com/s/wqn5v3wkktw9lmo/install.sh?dl=0 -O install.sh
sudo sh install.sh
python3 setup.py develop
# cp -R ${USER_DIR}/GermanData ../tmp/
python3 distribute.py --config_path config_tacotron_de.json --data_path ${USER_DIR}/GermanData/karlsson/
python3 distribute.py --config_path config_tacotron_de.json --data_path /data/rw/home/de_DE/
# cp -R ${USER_DIR}/Mozilla_22050 ../tmp/
# python3 distribute.py --config_path config_tacotron_gst.json --data_path ../tmp/Mozilla_22050/
while true; do sleep 1000000; done
52 changes: 8 additions & 44 deletions config_tacotron_de.json
View file
Original file line number Diff line number Diff line change
@@ -1,7 +1,6 @@
{
"github_branch":"tacotron-gst",
"run_name": "german-karlsson-tacotron-loc_attn",
"run_description": "train german with all of the german dataset",
"run_name": "german-all-tacotrongst",
"run_description": "train with all the german dataset using gst",

"audio":{
// Audio processing parameters
Expand Down Expand Up @@ -32,7 +31,7 @@

"reinit_layers": [],

"model": "Tacotron", // one of the model in models/
"model": "TacotronGST", // one of the model in models/
"grad_clip": 1, // upper limit for gradients for clipping.
"epochs": 10000, // total number of epochs to train.
"lr": 0.0001, // Initial learning rate. If Noam decay is active, maximum learning rate.
Expand All @@ -43,9 +42,9 @@
"attention_norm": "sigmoid", // softmax or sigmoid. Suggested to use softmax for Tacotron2 and sigmoid for Tacotron.
"prenet_type": "original", // ONLY TACOTRON2 - "original" or "bn".
"prenet_dropout": true, // ONLY TACOTRON2 - enable/disable dropout at prenet.
"use_forward_attn": true, // ONLY TACOTRON2 - if it uses forward attention. In general, it aligns faster.
"use_forward_attn": false, // ONLY TACOTRON2 - if it uses forward attention. In general, it aligns faster.
"transition_agent": false, // ONLY TACOTRON2 - enable/disable transition agent of forward attention.
"forward_attn_mask": true,
"forward_attn_mask": false,
"location_attn": true, // ONLY TACOTRON2 - enable_disable location sensitive attention. It is enabled for TACOTRON by default.
"loss_masking": true, // enable / disable loss masking against the sequence padding.
"enable_eos_bos_chars": false, // enable/disable beginning of sentence and end of sentence chars.
Expand All @@ -66,51 +65,16 @@
"test_sentences_file": "de_sentences.txt", // set a file to load sentences to be used for testing. If it is null then we use default english sentences.
"test_delay_epochs": 5, //Until attention is aligned, testing only wastes computation time.
"data_path": "/home/erogol/Data/m-ai-labs/de_DE/by_book/" , // DATASET-RELATED: can overwritten from command argument
"meta_file_train": [
// "/home/erogol/Data/m-ai-labs/de_DE/by_book/mix/erzaehlungen_poe/metadata.csv",
// "/home/erogol/Data/m-ai-labs/de_DE/by_book/mix/auf_zwei_planeten/metadata.csv",
"/home/erogol/Data/m-ai-labs/de_DE/by_book/male/karlsson/kleinzaches/metadata.csv",
"/home/erogol/Data/m-ai-labs/de_DE/by_book/male/karlsson/spiegel_kaetzchen/metadata.csv",
"/home/erogol/Data/m-ai-labs/de_DE/by_book/male/karlsson/herrnarnesschatz/metadata.csv",
"/home/erogol/Data/m-ai-labs/de_DE/by_book/male/karlsson/maedchen_von_moorhof/metadata.csv",
"/home/erogol/Data/m-ai-labs/de_DE/by_book/male/karlsson/koenigsgaukler/metadata.csv",
"/home/erogol/Data/m-ai-labs/de_DE/by_book/male/karlsson/altehous/metadata.csv",
"/home/erogol/Data/m-ai-labs/de_DE/by_book/male/karlsson/odysseus/metadata.csv",
"/home/erogol/Data/m-ai-labs/de_DE/by_book/male/karlsson/undine/metadata.csv",
"/home/erogol/Data/m-ai-labs/de_DE/by_book/male/karlsson/reise_tilsit/metadata.csv",
"/home/erogol/Data/m-ai-labs/de_DE/by_book/male/karlsson/schmied_seines_glueckes/metadata.csv",
"/home/erogol/Data/m-ai-labs/de_DE/by_book/male/karlsson/kammmacher/metadata.csv",
"/home/erogol/Data/m-ai-labs/de_DE/by_book/male/karlsson/unterm_birnbaum/metadata.csv",
"/home/erogol/Data/m-ai-labs/de_DE/by_book/male/karlsson/liebesbriefe/metadata.csv",
"/home/erogol/Data/m-ai-labs/de_DE/by_book/male/karlsson/sandmann/metadata.csv"
// "/home/erogol/Data/m-ai-labs/de_DE/by_book/female/eva_k/kleine_lord/metadata.csv",
// "/home/erogol/Data/m-ai-labs/de_DE/by_book/female/eva_k/toten_seelen/metadata.csv",
// "/home/erogol/Data/m-ai-labs/de_DE/by_book/female/eva_k/werde_die_du_bist/metadata.csv",
// "/home/erogol/Data/m-ai-labs/de_DE/by_book/female/eva_k/grune_haus/metadata.csv",
// "/home/erogol/Data/m-ai-labs/de_DE/by_book/female/rebecca_braunert_plunkett/das_letzte_marchen/metadata.csv",
// "/home/erogol/Data/m-ai-labs/de_DE/by_book/female/rebecca_braunert_plunkett/ferien_vom_ich/metadata.csv",
// "/home/erogol/Data/m-ai-labs/de_DE/by_book/female/rebecca_braunert_plunkett/maerchen/metadata.csv",
// "/home/erogol/Data/m-ai-labs/de_DE/by_book/female/rebecca_braunert_plunkett/mein_weg_als_deutscher_und_jude/metadata.csv",
// "/home/erogol/Data/m-ai-labs/de_DE/by_book/female/ramona_deininger/caspar/metadata.csv",
// "/home/erogol/Data/m-ai-labs/de_DE/by_book/female/ramona_deininger/sterben/metadata.csv",
// "/home/erogol/Data/m-ai-labs/de_DE/by_book/female/ramona_deininger/weihnachtsabend/metadata.csv",
// "/home/erogol/Data/m-ai-labs/de_DE/by_book/female/ramona_deininger/frankenstein/metadata.csv",
// "/home/erogol/Data/m-ai-labs/de_DE/by_book/female/ramona_deininger/tschun/metadata.csv",
// "/home/erogol/Data/m-ai-labs/de_DE/by_book/female/ramona_deininger/menschenhasser/metadata.csv",
// "/home/erogol/Data/m-ai-labs/de_DE/by_book/female/ramona_deininger/grune_gesicht/metadata.csv",
// "/home/erogol/Data/m-ai-labs/de_DE/by_book/female/ramona_deininger/tom_sawyer/metadata.csv",
// "/home/erogol/Data/m-ai-labs/de_DE/by_book/female/ramona_deininger/alter_afrikaner/metadata.csv",
// "/home/erogol/Data/m-ai-labs/de_DE/by_book/female/angela_merkel/merkel_alone/metadata.csv"
], // DATASET-RELATED: metafile for training dataloader.
"meta_file_train": null, // DATASET-RELATED: metafile for training dataloader.
"meta_file_val": null, // DATASET-RELATED: metafile for evaluation dataloader.
"dataset": "mailabs", // 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": 15, // DATASET-RELATED: minimum text length to use in training
"max_seq_len": 200, // DATASET-RELATED: maximum text length
"output_path": "/media/erogol/data_ssd/Data/models/mozilla_models/", // DATASET-RELATED: output path for all training outputs.
"output_path": "/home/erogol/Models/mozilla_models/", // DATASET-RELATED: output path for all training outputs.
"num_loader_workers": 4, // 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.
"phoneme_cache_path": "phoneme_cache", // 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": false, // use phonemes instead of raw characters. It is suggested for better pronounciation.
"phoneme_language": "de", // depending on your target language, pick one from https://github.com/bootphon/phonemizer#languages
"text_cleaner": "phoneme_cleaners"
}
Expand Down
18 changes: 11 additions & 7 deletions datasets/preprocess.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -62,20 +62,24 @@ def mozilla(root_path, meta_file):
def mailabs(root_path, meta_files):
"""Normalizes M-AI-Labs meta data files to TTS format"""
if meta_files is None:
meta_files = glob(root_path+"/**/metadata.csv", recursive=True)
folders = [f.strip().split("/")[-2] for f in meta_files]
csv_files = glob(root_path+"/**/metadata.csv", recursive=True)
folders = [os.path.dirname(f) for f in csv_files]
else:
folders = [f.strip().split("by_book")[1][1:] for f in meta_files]
csv_files = meta_files
folders = [f.strip().split("by_book")[1][1:] for f in csv_file]
# meta_files = [f.strip() for f in meta_files.split(",")]
items = []
for idx, meta_file in enumerate(meta_files):
print(" | > {}".format(meta_file))
for idx, csv_file in enumerate(csv_files):
print(" | > {}".format(csv_file))
folder = folders[idx]
txt_file = os.path.join(root_path, meta_file)
txt_file = os.path.join(root_path, csv_file)
with open(txt_file, 'r') as ttf:
for line in ttf:
cols = line.split('|')
wav_file = os.path.join(root_path, folder.replace("metadata.csv", ""), 'wavs', cols[0] + '.wav')
if meta_files is None:
wav_file = os.path.join(folder, 'wavs', cols[0] + '.wav')
else:
wav_file = os.path.join(root_path, folder.replace("metadata.csv", ""), 'wavs', cols[0] + '.wav')
if os.path.isfile(wav_file):
text = cols[1].strip()
items.append([text, wav_file])
Expand Down

0 comments on commit 118fe61

Please sign in to comment.