servinosmanov/crimean-tatar-tts-xtts-sevil

Crimean Tatar TTS - Sevil Voice (XTTS v2)

Text-to-speech synthesis for Crimean Tatar (Qırımtatar tili), a severely endangered Turkic language.

Overview

This Space demonstrates XTTS v2 model fine-tuned on Crimean Tatar speech data with the voice of native speaker Sevil.

Features:

• —High-quality speech synthesis
• —Support for both Latin and Cyrillic scripts
• —Adjustable temperature and speed parameters
• —Based on Coqui XTTS v2 architecture

Model

• —Model: servinosmanov/xtts-crh-sevil-v1
• —Dataset: MrSerWin/crimean-tatar-speech-sevil
• —Training data: ~2.5 hours of native speaker recordings
• —Architecture: XTTS v2 (GPT-based TTS)

Usage

Enter text in Crimean Tatar (Latin or Cyrillic script) and click "Generate" to synthesize speech.

Parameters

• —Temperature: Controls randomness in generation (0.1-1.0). Higher values produce more variation.
• —Speed: Speech speed multiplier (0.5-2.0).

Examples

• —Selâm! Nasılsıñız? - Hello! How are you?
• —Qırımtatar tili - güzel bir tildir. - Crimean Tatar is a beautiful language.
• —Menim adım Sevil. - My name is Sevil.

Related Resources

• —SpeechT5 Model - Alternative TTS model
• —Ana Yurt Lügat - Crimean Tatar Dictionary

License

MIT License

Author

Servin Osmanov

• —ORCID: 0009-0001-4010-0685
• —GitHub: @MrSerWin