pocket-tts
Kyutai's pocket-sized text-to-speech!
Decision gist · record as of 2026-08-14
Yes, if you need CPU-based TTS without GPU overhead and can accept unclear licensing. The package is actively maintained, installs easily, has no known vulnerabilities, and offers a practical balance of quality and speed. Verify the license terms before production use, and test performance on your target hardware. The 15 runtime dependencies are substantial but standard for ML workloads.AI-flagged interpretation of the facts on this page — verify before relying
Before you install
- Requires PyTorch 2.5+; Python 3.10–3.14 only.
- Model and voice state loading are slow operations; keep them in memory for repeated use.
- Low install friction with a pure-Python wheel.
License · maintenance · safety
(unclear) — License treatment is unclear; no SPDX identifier or raw license text is available in the metadata. Verify the actual license before use in proprietary or copyleft-sensitive projects.
last release 2026-05-04 (102 days)
0 known vulnerabilities (OSV.dev, 2026-08-14) · 76,294 downloads/mo, #14,639 on PyPI
Alternatives
Verify before relying
pip install pocket-tts
from pocket_tts import TTSModel
import scipy.io.wavfile
tts_model = TTSModel.load_model()
voice_state = tts_model.get_state_for_audio_prompt("alba")
audio = tts_model.generate_audio(voice_state, "Hello world")
scipy.io.wavfile.write("output.wav", tts_model.sample_rate, audio.numpy())- Actual license terms and attribution requirements for the model and voice samples
- Whether voice cloning respects speaker consent and licensing of source audio
- Performance characteristics on systems other than MacBook Air M4
- Stability and API compatibility across minor versions
What it is and what it does
Pocket TTS is a CPU-based text-to-speech engine built on a 100M-parameter model that generates natural speech without requiring GPU hardware or external APIs. It runs on Python 3.10–3.14 with PyTorch 2.5+ and provides both a command-line interface and a Python library for integration into applications. The model supports six languages (English, French, German, Portuguese, Italian, Spanish) and includes voice cloning from audio samples, streaming output, and a local HTTP server for batch or interactive use.
The package depends on a substantial ML stack—torch, numpy, scipy, einops, safetensors, huggingface-hub—plus FastAPI and Uvicorn for the server mode. It achieves low-latency streaming (first chunk in ~200ms) and runs faster than real-time on modest CPU hardware. The library is designed for straightforward integration: load the model once, create voice states for each speaker, then generate audio in a single function call. Voice cloning requires preprocessing audio files into safetensors embeddings for fast inference.
Use it for
- Generate speech from user input in a web or desktop application without deploying a GPU server or calling a cloud API.
- Clone a speaker's voice from a sample audio file and synthesize new utterances in that voice for accessibility or personalization.
- Build a local TTS microservice with the `serve` command to batch-process text-to-speech requests over HTTP.
- Integrate multilingual speech synthesis into a Python script or Jupyter notebook for data annotation, testing, or prototyping.
- Stream audio output in real time to reduce latency when generating long or interactive speech content.
Worth the install?
AI-flagged interpretation of the facts on this page. Verify before relying on it.
Yes, if you need CPU-based TTS without GPU overhead and can accept unclear licensing.
The package is actively maintained, installs easily, has no known vulnerabilities, and offers a practical balance of quality and speed. Verify the license terms before production use, and test performance on your target hardware. The 15 runtime dependencies are substantial but standard for ML workloads.
Install
pocket-tts on PyPI
Before you install
Low install friction with a pure-Python wheel. Active maintenance as of 102 days since last release. Requires PyTorch 2.5+ and Python 3.10–3.14, but does not mandate GPU PyTorch.
Requires PyTorch 2.5+; Python 3.10–3.14 only. Model and voice state loading are slow operations; keep them in memory for repeated use.
License in practice
License treatment is unclear; no SPDX identifier or raw license text is available in the metadata. Verify the actual license before use in proprietary or copyleft-sensitive projects.
Quickstart
pip install pocket-tts
from pocket_tts import TTSModel
import scipy.io.wavfile
tts_model = TTSModel.load_model()
voice_state = tts_model.get_state_for_audio_prompt("alba")
audio = tts_model.generate_audio(voice_state, "Hello world")
scipy.io.wavfile.write("output.wav", tts_model.sample_rate, audio.numpy())Verify before relying
- Actual license terms and attribution requirements for the model and voice samples
- Whether voice cloning respects speaker consent and licensing of source audio
- Performance characteristics on systems other than MacBook Air M4
- Stability and API compatibility across minor versions
Package facts
| License | Not declared unclear |
| Python support | Supports the current Python release <3.15,>=3.10 |
| Install friction | Low. Pure-Python wheel |
| Runtime dependencies | 15 packagesbeartypeeinopsfastapihuggingface-hubnumpypydanticpython-multipartrequestssafetensorsscipysentencepiecetorchtypertyping-extensionsuvicorn |
| Maintenance | Actively maintained 102 days since the last release |
| First released | |
| Downloads | 76,294 / month, #14,639 on PyPI 30-day window, as of 2026-08-14 |
| Known vulnerabilities | None known OSV.dev, checked 2026-08-14 |
Evidence: pocket_tts-2.1.0-py3-none-any.whl
Tags
Let your AI agent find packages like this
Example. Real query, live index.
You found this page by searching. An agent finds it by wishing: SkillFed indexes 14,416 PyPI packages by what they can do, searchable in plain language.
wish › “text to speech cpu”
- pocket-ttsPocket TTS generates speech from text on CPU using a lightweight…
- faster-whisperTranscribes audio to text using OpenAI's Whisper model, reimplemented…
- sileroSilero provides pre-trained text-to-speech models that convert text…
Give your agent the search over MCP, or paste the wish link into any chat.
More Artificial Intelligence packages
LiteLLM provides a unified Python interface to call 100+ LLM providers (OpenAI, Anthropic, Gemini, Bedrock, Azure, and others) using OpenAI-compatible API format, available as both a Python SDK and a self-hosted AI Gateway proxy server.
Install it if you need to work with multiple LLM providers or want to centralize LLM routing in your organization.
Client library and CLI tool for downloading, uploading, and managing models, datasets, and repositories on the Hugging Face Hub platform.
Install it if you work with Hugging Face Hub models or datasets.
LangChain provides a framework for building agents and LLM-powered applications by composing language models, tools, and memory through a unified API that abstracts over multiple model providers.
hf-xet provides chunk-based deduplication and efficient file transfer for the Hugging Face Hub, enabling faster uploads and downloads of large files with local disk caching.
Tokenizers converts raw text into token sequences for NLP models, with support for training custom vocabularies and using pre-built tokenizers (BPE, WordPiece) optimized for speed via Rust.
Transformers provides a unified framework for loading, fine-tuning, and running state-of-the-art pretrained models across text, vision, audio, video, and multimodal tasks using PyTorch, JAX, or TensorFlow.
Install it if you need to run or train any transformer-based model for NLP, vision, audio, or multimodal tasks.
See also TTS · pyttsx3 · piper-tts · kokoro-onnx · chatterbox-tts · voxcpm · gTTS · omnivoice · qwen-tts · edge-tts