gabriele-mastrapasqua/qwen3-tts

Pure C inference engine for Qwen3-TTS text-to-speech. No Python, no PyTorch — just C and BLAS. Supports 0.6B and 1.7B models, 9 voices, 10 languages.

45
/ 100
Emerging

Implements the complete TTS pipeline directly in C: BPE tokenization, a 28-layer causal transformer decoder, multi-pass code prediction, and convolutional speech synthesis—all with BF16 weights memory-mapped from safetensors for near-instant loading. Includes voice cloning from reference audio, style control via instruction prompts, INT8/INT4 quantization for speedup, and an OpenAI-compatible HTTP server (`/v1/audio/speech`, `/v1/tts/stream`) for integration into larger applications.

No Package No Dependents
Maintenance 13 / 25
Adoption 7 / 25
Maturity 9 / 25
Community 16 / 25

How are scores calculated?

Stars

25

Forks

6

Language

C

License

MIT

Last pushed

Mar 16, 2026

Commits (30d)

0

Get this data via API

curl "https://pt-edge.onrender.com/api/v1/quality/voice-ai/gabriele-mastrapasqua/qwen3-tts"

Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.