Sundanese Fine-Tuned Qwen3-TTS
Fine-tuned Qwen3-TTS 0.6B for Sundanese (Sunda) voice synthesis using OpenSLR SLR44 dataset.
Model Details
- Base Model:
Qwen/Qwen3-TTS-12Hz-0.6B-Base - Language: Sundanese (su)
- Dataset: OpenSLR SLR44 - Female voice only
- Source: Mozilla Common Voice - Sundanese Female
- Training: 4 epochs, lr=1e-6, batch_size=2
- Speaker: sunda_speaker
Usage
from qwen_tts import Qwen3TTSModel
import soundfile as sf
model = Qwen3TTSModel.from_pretrained("alkhrzmy/qwen3-tts-0.6b-indonesian")
wavs, sr = model.generate_custom_voice(
text="Selamat pagi, ini uji coba suara Indonesia.",
language="auto",
speaker="sunda_speaker"
)
sf.write("output.wav", wavs[0], sr)
Training Data Citation
@inproceedings{kjartansson-etal-tts-sltu2018,
title = {{A Step-by-Step Process for Building TTS Voices Using Open Source Data and Framework for Bangla, Javanese, Khmer, Nepali, Sinhala, and Sundanese}},
author = {Keshan Sodimana and Knot Pipatsrisawat and Linne Ha and Martin Jansche and Oddur Kjartansson and Pasindu De Silva and Supheakmungkol Sarin},
booktitle = {Proc. The 6th Intl. Workshop on Spoken Language Technologies for Under-Resourced Languages (SLTU)},
year = {2018},
address = {Gurugram, India},
month = aug,
pages = {66--70},
URL = {http://dx.doi.org/10.21437/SLTU.2018-14}
}