README.md

metadata

library_name: transformers
datasets:
  - malaysia-ai/Multilingual-TTS
  - Scicom-intl/Emilia-YODAS-Voice-Conversion
  - Scicom-intl/Malaysian-Emilia
base_model:
  - Qwen/Qwen3-4B-Base
language:
  - en
  - ms
  - zh
  - ta

Multilingual-TTS-4B-Base

Continue pretraining Qwen/Qwen3-4B-Base on Multilingual Voice Conversion and TTS.

1. Use neucodec as speech detokenizer, 50 TPS, output in 24k sample rate.
2. Multi-speaker multilingual Voice Conversion, up to 35.88B tokens.
3. Multi-speaker multilingual TTS more than 150 languages, up to 14.64B tokens.
4. Flash Attention 3 10k context length varlen multipacking.
5. BF16 training.
6. MuonAdamW optimizer.