cat > README.md << 'EOF'

language:

• kk tags:
• text-to-speech
• tts
• kazakh
• qwen3-tts
• emotional-speech license: cc-by-4.0 datasets:
• issai/KazEmoTTS base_model: Qwen/Qwen3-TTS-12Hz-1.7B-Base

---

Qwen3-TTS Kazakh Emotional Speech

Fine-tuned Qwen3-TTS model for Kazakh language text-to-speech with emotional expression.

Training Data

• Dataset: KazEmoTTS (Kazakh Emotional TTS)
• Speaker: Female (kaz_aru) - 34 hours
• Emotions: neutral, happy, sad, angry, fear, surprise

Usage

import torch
from qwen_tts import Qwen3TTSModel

model = Qwen3TTSModel.from_pretrained(
    "futureDoctor/qwen3-tts-kazakh-emotional",
    device_map="cuda:0",
    dtype=torch.bfloat16,
)

wavs, sr = model.generate_custom_voice(
    text="Сәлеметсіз бе! Мен сізге көмектесемін.",
    speaker="kazakh_female_emotional",
)