kyutai-labs/pocket-tts

Health v2 · maintenance only

Adoption not part of the score

9135 stars · 940 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-29, confidence not recorded

Pocket TTS is a lightweight 100M-parameter text-to-speech model and Python library from Kyutai that runs efficiently on CPUs without a GPU. It offers a Python API and CLI with streaming audio, low latency, voice cloning, and multi-language support.

Use cases

• generate speech from text on a CPU
• convert text to wav audio files from the command line
• clone a voice for text-to-speech
• stream synthesized speech with low latency
• run TTS locally without a GPU or web API
• generate speech in multiple languages like French or German
• run text-to-speech in the browser client-side
• train a custom TTS model

When to choose

• you need fast, real-time TTS on CPU-only hardware
• you want local speech synthesis without cloud API costs
• you need voice cloning or multilingual speech generation
• you want a simple pip-installable TTS with a Python API and CLI

When to avoid

• you need the highest possible speech quality from large GPU-based models
• you need languages beyond the currently supported set
• you require non-Python integration without a serving layer

Facets

library · maturity active

tts audio-processing machine-learning cli llm-inference speech-processing machine-learning artificial-intelligence python cross-platform cli browser windows voice-cloning cpu-inference streaming-audio multilingual text-to-speech audio macos linux

Member repositories

For agents

markdown · JSON · MCP: product_card(name="kyutai-labs/pocket-tts")