Piper TTS
Fast offline neural text-to-speech that runs on any CPU.
About
Piper TTS is an open-source neural text-to-speech engine created by the Rhasspy project, now maintained under the Open Home Foundation. It converts text into natural-sounding speech using compact ONNX voice models that run entirely on-device, with no cloud API calls and no internet connection required.
The engine uses VITS-based neural models paired with an espeak-ng phonemizer, and it generates speech faster than real time even on modest hardware such as a Raspberry Pi. Pre-trained voices cover 30+ languages, and the project is widely embedded in home assistants, accessibility tools, and self-hosted dashboards where sending text to third-party servers is not acceptable.
What it can do
- Offline neural speech synthesis with no cloud dependency
- 30+ languages with compact ONNX voice models
- Faster-than-real-time generation on CPU
- CLI, C++ library, and Python bindings
- Streaming audio output for low-latency playback
Setup
Install the piper-tts Python package or download a prebuilt binary, then fetch a voice model (.onnx + .json) from the Hugging Face voice repository.
Prerequisites
A CPU is enough; no GPU needed. Voice model files of 15-100MB each.
Use cases
Works with
Trust score — 8/10
Scored 8/10 (verified tier, metrics-v1). Evidence: Verified via vendor site/docs/GitHub and reputable coverage, wave 125; Trending score 78/100 from researcher assessment.
Sources
Every fact above was checked against these sources. Where a claim comes from the provider and couldn't be independently verified, the text says so.