- OS: Ubuntu 24.04.2
- CUDA Toolkit: 11.7
- GPU Driver: NVIDIA-SMI 570.144 (CUDA Version 12.8)
1-1. Create ap_env
environment
cd /IIPL_Flitto/AdaptiVoice/TTS_engine
conda create -n ap_env python=3.9
conda activate ap_env
1-2. Install packages
pip install -e .
cd /IIPL_Flitto/AdaptiVoice/voice_engine
pip install -e .
1-3. Install additional packages
conda install -c conda-forge ffmpeg
pip install huggingface_hub==0.14.0
pip install mecab-python3
python -m unidic download
conda install -c conda-forge gxx_linux-64
pip install pkuseg janome konlpy h5py textgrid tgt opencc librosa
2-1. Create mfa_env
environment
conda create -n mfa_env -c conda-forge montreal-forced-aligner
conda activate mfa_env
2-2. Install packages
pip install joblib==1.2.0
pip install python-mecab-ko jamo spacy-pkuseg dragonmapper hanziconv textgrid tgt
conda install -c conda-forge spacy sudachipy sudachidict-core
Download the Crossview_AP Model checkpoint.
- config 파일 경로 수정
- IIPL_Flitto/metric/Crossview-AP/data/config/config_en.json
- 각 언어 지정 필요
bash /IIPL_Flitto/TTA_test/crossview_ap.shcrossview_ap.sh
- TTS
- MFA