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

Rename ununsed_speakers to ignored_speakers

This commit is contained in:
Edresson 2021-11-22 18:42:54 -03:00 committed by Eren Gölge
parent 346bd931b2
commit aa1a070d58
3 changed files with 25 additions and 26 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
TTS/config/shared_configs.py
View File

@ -198,7 +198,7 @@ class BaseDatasetConfig(Coqpit):
name: str = "" name: str = ""
path: str = "" path: str = ""
meta_file_train: str = "" meta_file_train: str = ""
ununsed_speakers: List[str] = None ignored_speakers: List[str] = None
language: str = "" language: str = ""
meta_file_val: str = "" meta_file_val: str = ""
meta_file_attn_mask: str = "" meta_file_attn_mask: str = ""

7
TTS/tts/datasets/__init__.py
View File

@ -67,22 +67,21 @@ def load_tts_samples(
root_path = dataset["path"] root_path = dataset["path"]
meta_file_train = dataset["meta_file_train"] meta_file_train = dataset["meta_file_train"]
meta_file_val = dataset["meta_file_val"] meta_file_val = dataset["meta_file_val"]
ununsed_speakers = dataset["ununsed_speakers"] ignored_speakers = dataset["ignored_speakers"]
language = dataset["language"] language = dataset["language"]
# setup the right data processor # setup the right data processor
if formatter is None: if formatter is None:
formatter = _get_formatter_by_name(name) formatter = _get_formatter_by_name(name)
# load train set # load train set
meta_data_train = formatter(root_path, meta_file_train, ununsed_speakers=ununsed_speakers) meta_data_train = formatter(root_path, meta_file_train, ignored_speakers=ignored_speakers)
# TODO: remove the loops and pass language as a parameter to preprocessor for faster load
meta_data_train = [[*item, language] for item in meta_data_train] meta_data_train = [[*item, language] for item in meta_data_train]
print(f" | > Found {len(meta_data_train)} files in {Path(root_path).resolve()}") print(f" | > Found {len(meta_data_train)} files in {Path(root_path).resolve()}")
# load evaluation split if set # load evaluation split if set
if eval_split: if eval_split:
if meta_file_val: if meta_file_val:
meta_data_eval = formatter(root_path, meta_file_val, ununsed_speakers=ununsed_speakers) meta_data_eval = formatter(root_path, meta_file_val, ignored_speakers=ignored_speakers)
meta_data_eval = [[*item, language] for item in meta_data_eval] meta_data_eval = [[*item, language] for item in meta_data_eval]
else: else:
meta_data_eval, meta_data_train = split_dataset(meta_data_train) meta_data_eval, meta_data_train = split_dataset(meta_data_train)

42
TTS/tts/datasets/formatters.py
View File

@ -59,7 +59,7 @@ def mozilla_de(root_path, meta_file, **kwargs): # pylint: disable=unused-argume
return items return items
def mailabs(root_path, meta_files=None, ununsed_speakers=None): def mailabs(root_path, meta_files=None, ignored_speakers=None):
"""Normalizes M-AI-Labs meta data files to TTS format """Normalizes M-AI-Labs meta data files to TTS format
Args: Args:
@ -88,8 +88,8 @@ def mailabs(root_path, meta_files=None, ununsed_speakers=None):
continue continue
speaker_name = speaker_name_match.group("speaker_name") speaker_name = speaker_name_match.group("speaker_name")
# ignore speakers # ignore speakers
if isinstance(ununsed_speakers, list): if isinstance(ignored_speakers, list):
if speaker_name in ununsed_speakers: if speaker_name in ignored_speakers:
continue continue
print(" | > {}".format(csv_file)) print(" | > {}".format(csv_file))
with open(txt_file, "r", encoding="utf-8") as ttf: with open(txt_file, "r", encoding="utf-8") as ttf:
@ -197,7 +197,7 @@ def nancy(root_path, meta_file, **kwargs): # pylint: disable=unused-argument
return items return items
def common_voice(root_path, meta_file, ununsed_speakers=None): def common_voice(root_path, meta_file, ignored_speakers=None):
"""Normalize the common voice meta data file to TTS format.""" """Normalize the common voice meta data file to TTS format."""
txt_file = os.path.join(root_path, meta_file) txt_file = os.path.join(root_path, meta_file)
items = [] items = []
@ -209,15 +209,15 @@ def common_voice(root_path, meta_file, ununsed_speakers=None):
text = cols[2] text = cols[2]
speaker_name = cols[0] speaker_name = cols[0]
# ignore speakers # ignore speakers
if isinstance(ununsed_speakers, list): if isinstance(ignored_speakers, list):
if speaker_name in ununsed_speakers: if speaker_name in ignored_speakers:
continue continue
wav_file = os.path.join(root_path, "clips", cols[1].replace(".mp3", ".wav")) wav_file = os.path.join(root_path, "clips", cols[1].replace(".mp3", ".wav"))
items.append([text, wav_file, "MCV_" + speaker_name]) items.append([text, wav_file, "MCV_" + speaker_name])
return items return items
def libri_tts(root_path, meta_files=None, ununsed_speakers=None): def libri_tts(root_path, meta_files=None, ignored_speakers=None):
"""https://ai.google/tools/datasets/libri-tts/""" """https://ai.google/tools/datasets/libri-tts/"""
items = [] items = []
if not meta_files: if not meta_files:
@ -237,8 +237,8 @@ def libri_tts(root_path, meta_files=None, ununsed_speakers=None):
wav_file = os.path.join(_root_path, file_name + ".wav") wav_file = os.path.join(_root_path, file_name + ".wav")
text = cols[2] text = cols[2]
# ignore speakers # ignore speakers
if isinstance(ununsed_speakers, list): if isinstance(ignored_speakers, list):
if speaker_name in ununsed_speakers: if speaker_name in ignored_speakers:
continue continue
items.append([text, wav_file, "LTTS_" + speaker_name]) items.append([text, wav_file, "LTTS_" + speaker_name])
for item in items: for item in items:
@ -265,7 +265,7 @@ def custom_turkish(root_path, meta_file, **kwargs): # pylint: disable=unused-ar
# ToDo: add the dataset link when the dataset is released publicly # ToDo: add the dataset link when the dataset is released publicly
def brspeech(root_path, meta_file, ununsed_speakers=None): def brspeech(root_path, meta_file, ignored_speakers=None):
"""BRSpeech 3.0 beta""" """BRSpeech 3.0 beta"""
txt_file = os.path.join(root_path, meta_file) txt_file = os.path.join(root_path, meta_file)
items = [] items = []
@ -278,14 +278,14 @@ def brspeech(root_path, meta_file, ununsed_speakers=None):
text = cols[2] text = cols[2]
speaker_id = cols[3] speaker_id = cols[3]
# ignore speakers # ignore speakers
if isinstance(ununsed_speakers, list): if isinstance(ignored_speakers, list):
if speaker_id in ununsed_speakers: if speaker_id in ignored_speakers:
continue continue
items.append([text, wav_file, speaker_id]) items.append([text, wav_file, speaker_id])
return items return items
def vctk(root_path, meta_files=None, wavs_path="wav48", ununsed_speakers=None): def vctk(root_path, meta_files=None, wavs_path="wav48", ignored_speakers=None):
"""homepages.inf.ed.ac.uk/jyamagis/release/VCTK-Corpus.tar.gz""" """homepages.inf.ed.ac.uk/jyamagis/release/VCTK-Corpus.tar.gz"""
items = [] items = []
meta_files = glob(f"{os.path.join(root_path,'txt')}/**/*.txt", recursive=True) meta_files = glob(f"{os.path.join(root_path,'txt')}/**/*.txt", recursive=True)
@ -293,8 +293,8 @@ def vctk(root_path, meta_files=None, wavs_path="wav48", ununsed_speakers=None):
_, speaker_id, txt_file = os.path.relpath(meta_file, root_path).split(os.sep) _, speaker_id, txt_file = os.path.relpath(meta_file, root_path).split(os.sep)
file_id = txt_file.split(".")[0] file_id = txt_file.split(".")[0]
# ignore speakers # ignore speakers
if isinstance(ununsed_speakers, list): if isinstance(ignored_speakers, list):
if speaker_id in ununsed_speakers: if speaker_id in ignored_speakers:
continue continue
with open(meta_file, "r", encoding="utf-8") as file_text: with open(meta_file, "r", encoding="utf-8") as file_text:
text = file_text.readlines()[0] text = file_text.readlines()[0]
@ -304,7 +304,7 @@ def vctk(root_path, meta_files=None, wavs_path="wav48", ununsed_speakers=None):
return items return items
def vctk_slim(root_path, meta_files=None, wavs_path="wav48", ununsed_speakers=None): # pylint: disable=unused-argument def vctk_slim(root_path, meta_files=None, wavs_path="wav48", ignored_speakers=None): # pylint: disable=unused-argument
"""homepages.inf.ed.ac.uk/jyamagis/release/VCTK-Corpus.tar.gz""" """homepages.inf.ed.ac.uk/jyamagis/release/VCTK-Corpus.tar.gz"""
items = [] items = []
txt_files = glob(f"{os.path.join(root_path,'txt')}/**/*.txt", recursive=True) txt_files = glob(f"{os.path.join(root_path,'txt')}/**/*.txt", recursive=True)
@ -312,8 +312,8 @@ def vctk_slim(root_path, meta_files=None, wavs_path="wav48", ununsed_speakers=No
_, speaker_id, txt_file = os.path.relpath(text_file, root_path).split(os.sep) _, speaker_id, txt_file = os.path.relpath(text_file, root_path).split(os.sep)
file_id = txt_file.split(".")[0] file_id = txt_file.split(".")[0]
# ignore speakers # ignore speakers
if isinstance(ununsed_speakers, list): if isinstance(ignored_speakers, list):
if speaker_id in ununsed_speakers: if speaker_id in ignored_speakers:
continue continue
wav_file = os.path.join(root_path, wavs_path, speaker_id, file_id + ".wav") wav_file = os.path.join(root_path, wavs_path, speaker_id, file_id + ".wav")
items.append([None, wav_file, "VCTK_" + speaker_id]) items.append([None, wav_file, "VCTK_" + speaker_id])
@ -321,7 +321,7 @@ def vctk_slim(root_path, meta_files=None, wavs_path="wav48", ununsed_speakers=No
return items return items
def mls(root_path, meta_files=None, ununsed_speakers=None): def mls(root_path, meta_files=None, ignored_speakers=None):
"""http://www.openslr.org/94/""" """http://www.openslr.org/94/"""
items = [] items = []
with open(os.path.join(root_path, meta_files), "r", encoding="utf-8") as meta: with open(os.path.join(root_path, meta_files), "r", encoding="utf-8") as meta:
@ -331,8 +331,8 @@ def mls(root_path, meta_files=None, ununsed_speakers=None):
speaker, book, *_ = file.split("_") speaker, book, *_ = file.split("_")
wav_file = os.path.join(root_path, os.path.dirname(meta_files), "audio", speaker, book, file + ".wav") wav_file = os.path.join(root_path, os.path.dirname(meta_files), "audio", speaker, book, file + ".wav")
# ignore speakers # ignore speakers
if isinstance(ununsed_speakers, list): if isinstance(ignored_speakers, list):
if speaker in ununsed_speakers: if speaker in ignored_speakers:
continue continue
items.append([text, wav_file, "MLS_" + speaker]) items.append([text, wav_file, "MLS_" + speaker])
return items return items