So first I download the project
- So I use qwen3:14B and use whisper small set up kobold
- I go to in Yaml and remove all the cloud and keep the local like kobold, melotts (remove kobold tts)
- Then I download:
https://github.com/manhcuong02/MeloTTS_Vietnamese
- And then I put it in model/melotts
- Next up I go to this and download in file pretrain (download one file with the config): https://huggingface.co/nmcuong/MeloTTS-Vietnamese/tree/main
- then open power shell to download pip install underthesea, segments
- And change a bit in the test_infer.ipynb like the Run Inference tell at step 3
then run test with 2 language EN and VI
code for En:
python -c "from melo.api import TTS; model=TTS(language='EN', device='cuda'); speaker_ids=model.hps.data.spk2id; model.tts_to_file('Hello, this is a MeloTTS test.', speaker_ids['EN-Default'], 'test.wav'); print('done')"
code for vn:
python -c "from melo.api import TTS; model=TTS(language='VI', config_path='models/config.json', ckpt_path='models/G_463000.pth', device='cuda'); model.tts_to_file('Xin chào bạn. Đây là đoạn kiểm tra tiếng Việt.', 0, r'D:\jaison-core-2.2.1\models\melotts\MeloTTS_Vietnamese\vi_test.wav'); print('DONE')"
P/s: it might miss some model so you need to download some for it to run
- Then I move both of it to the melotts-vietnamese and make a new folder for it
- After that I download RVC and put it in model/rvc
- then set up on env-plate to env and do like what readme tell
- Then download ffmpeg like the readme
- And finally start with python ./src/main.py --config=example
But it end up like this
PS D:\jaison-core-2.2.1> python ./src/main.py --config=example
[2026-05-15 22:10:36,759] [INFO ] [jaison.py::83 start]: Starting JAIson application layer.
[2026-05-15 22:10:36,949] [INFO ] [operation.py::32 start]: Starting STT operation kobold
[2026-05-15 22:10:36,949] [INFO ] [manager.py::23 load]: Loading process by type kobold
[2026-05-15 22:10:36,971] [INFO ] [koboldcpp.py::32 reload]: Opened Koboldcpp server (PID: 4872) on port 61977
[2026-05-15 22:10:36,975] [INFO ] [operation.py::32 start]: Starting T2T operation kobold
[2026-05-15 22:10:36,991] [INFO ] [operation.py::32 start]: Starting FILTER_TEXT operation filter_clean
PS D:\jaison-core-2.2.1>
pls help
So first I download the project
https://github.com/manhcuong02/MeloTTS_Vietnamese
then run test with 2 language EN and VI
code for En:
python -c "from melo.api import TTS; model=TTS(language='EN', device='cuda'); speaker_ids=model.hps.data.spk2id; model.tts_to_file('Hello, this is a MeloTTS test.', speaker_ids['EN-Default'], 'test.wav'); print('done')"
code for vn:
python -c "from melo.api import TTS; model=TTS(language='VI', config_path='models/config.json', ckpt_path='models/G_463000.pth', device='cuda'); model.tts_to_file('Xin chào bạn. Đây là đoạn kiểm tra tiếng Việt.', 0, r'D:\jaison-core-2.2.1\models\melotts\MeloTTS_Vietnamese\vi_test.wav'); print('DONE')"
P/s: it might miss some model so you need to download some for it to run
But it end up like this
PS D:\jaison-core-2.2.1> python ./src/main.py --config=example
[2026-05-15 22:10:36,759] [INFO ] [jaison.py::83 start]: Starting JAIson application layer.
[2026-05-15 22:10:36,949] [INFO ] [operation.py::32 start]: Starting STT operation kobold
[2026-05-15 22:10:36,949] [INFO ] [manager.py::23 load]: Loading process by type kobold
[2026-05-15 22:10:36,971] [INFO ] [koboldcpp.py::32 reload]: Opened Koboldcpp server (PID: 4872) on port 61977
[2026-05-15 22:10:36,975] [INFO ] [operation.py::32 start]: Starting T2T operation kobold
[2026-05-15 22:10:36,991] [INFO ] [operation.py::32 start]: Starting FILTER_TEXT operation filter_clean
PS D:\jaison-core-2.2.1>
pls help