Sanket/ai-voice-cloning
1
0
Fork 0
forked from mrq/ai-voice-cloning
Code Pull Requests Activity

fixed setup scripts and Dockerfile to NOT use extra-index-url and instead use index-url (how this happened I don't know, since pytorch instructions use index-url), '''fixed''' phonemizing japanese for VALL-E with pykakasi

Browse Source
This commit is contained in:
mrq 2023-10-12 00:27:46 +00:00
parent 17acfee5d0
commit 2830d1fa96
6 changed files with 26 additions and 5 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
Dockerfile Normal file → Executable file
View File

@ -20,7 +20,7 @@ ENV PATH="$HOME/miniconda/bin:$PATH"
RUN conda init
RUN conda install python=$PYTHON_VERSION
RUN python3 -m pip install --upgrade pip
RUN pip3 install torch torchvision torchaudio --extra-index-url https://download.pytorch.org/whl/cu118
RUN pip3 install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu118
RUN mkdir $HOME/ai-voice-cloning
WORKDIR $HOME/ai-voice-cloning

1
requirements.txt
View File

@ -10,3 +10,4 @@ phonemizer
pydantic==1.10.11
websockets
beartype==0.15.0
pykakasi

2
setup-cuda.bat
View File

@ -4,7 +4,7 @@ git submodule update --remote
python -m venv venv
call .\venv\Scripts\activate.bat
python -m pip install --upgrade pip
python -m pip install torch torchvision torchaudio --extra-index-url https://download.pytorch.org/whl/cu118
python -m pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu118
python -m pip install -r .\modules\tortoise-tts\requirements.txt
python -m pip install -e .\modules\tortoise-tts\
python -m pip install -r .\modules\dlas\requirements.txt

2
setup-cuda.sh
View File

@ -7,7 +7,7 @@ python3 -m venv venv
source ./venv/bin/activate
python3 -m pip install --upgrade pip # just to be safe
# CUDA
pip3 install torch torchvision torchaudio --extra-index-url https://download.pytorch.org/whl/cu118
pip3 install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu118
# install requirements
python3 -m pip install -r ./modules/tortoise-tts/requirements.txt # install TorToiSe requirements
python3 -m pip install -e ./modules/tortoise-tts/ # install TorToiSe

2
setup-rocm.sh
View File

@ -7,7 +7,7 @@ python3 -m venv venv
source ./venv/bin/activate
python3 -m pip install --upgrade pip # just to be safe
# ROCM
pip3 install torch==1.13.1 torchvision torchaudio --extra-index-url https://download.pytorch.org/whl/rocm5.2 # 5.4.2 doesn't work for me desu
pip3 install torch==1.13.1 torchvision torchaudio --index-url https://download.pytorch.org/whl/rocm5.2 # 5.4.2 doesn't work for me desu
# install requirements
python3 -m pip install -r ./modules/tortoise-tts/requirements.txt # install TorToiSe requirements
python3 -m pip install -e ./modules/tortoise-tts/ # install TorToiSe

20
src/utils.py
View File

@ -68,8 +68,20 @@ BARK_ENABLED = False
VERBOSE_DEBUG = True
KKS = None
PYKAKASI_ENABLED = False
import traceback
try:
import pykakasi
KKS = pykakasi.kakasi()
PYKAKASI_ENABLED = True
except Exception as e:
#if VERBOSE_DEBUG:
# print(traceback.format_exc())
pass
try:
from whisper.normalizers.english import EnglishTextNormalizer
from whisper.normalizers.basic import BasicTextNormalizer
@ -2734,6 +2746,14 @@ def prepare_dataset( voice, use_segments=False, text_length=0, audio_length=0, p
phn_file = jobs['phonemize'][0][i]
normalized = jobs['phonemize'][1][i]
if language == "japanese":
language = "ja"
if language == "ja" and PYKAKASI_ENABLED and KKS is not None:
normalized = KKS.convert(normalized)
normalized = [ n["hira"] for n in normalized ]
normalized = "".join(normalized)
try:
phonemized = valle_phonemize( normalized )
open(phn_file, 'w', encoding='utf-8').write(" ".join(phonemized))