LuxTTS vs pocket-tts
Lightweight voice-cloning TTS — 48kHz speech at 150x realtime, fits in 1GB VRAM and runs on CPU or MPS. SOTA cloning from a ~3s reference sample, rivaling models 10x larger. — versus — Kyutai's 100M-parameter CPU-only TTS — streaming audio in ~200ms, ~6× real-time on two laptop cores, voice cloning, six languages. pip install and it talks.
Both are tiny local TTS engines: pocket-tts is Kyutai's 100M CPU-only streaming model with a fixed voice set; LuxTTS targets high-fidelity 48kHz voice cloning from a reference sample. Pick by whether you need cloning or streaming.
| LuxTTS | pocket-tts | |
|---|---|---|
| Stars | 4.8k | 7.6k |
| Forks | 623 | 776 |
| Language | Python | Python |
| License | Apache-2.0 | MIT |
| Last activity | 1 months ago | 23 days ago |
| Topics | voice, local | local, voice |
| Curated connections | 2 | 3 |
LuxTTS — the curator's take
Reach for it when you need fast, local voice cloning that fits anywhere: 48kHz output (most open TTS caps at 24kHz), sub-1GB VRAM, and faster-than-realtime even on CPU make it practical for on-device apps and batch narration. NOT for you if you need many built-in voices or languages out of the box — it clones a reference, it doesn't ship a voice library — and cloning any real person's voice without consent is an ethics/legal minefield. Quality rides on a clean 3s+ reference.
pocket-tts — the curator's take
This is the TTS you add when the rest of your stack is already local: no GPU, no API key, ~200ms to first audio on two CPU cores, and a port ecosystem (WASM, MLX, ONNX, C++, Home Assistant) that means it runs basically anywhere — the natural voice for an Ollama-powered assistant. When NOT: you need studio-grade expressiveness or high-concurrency serving — it's a batch-of-1, small-model design, not a production voice API; and the polished voice catalog is English-heaviest. Voice cloning needs 20s of audio — use it with consent.