Piper
Fast, local neural text-to-speech — archived upstream, continued as piper1-gpl.
Retired
Piper appears to be discontinued or shut down — rhasspy/piper archived Aug 2025; dev moved to OHF-Voice/piper1-gpl This page is kept as an archive record and is excluded from search engines.
About
Piper is the beloved local neural text-to-speech engine: fast, fully offline, private, with 100+ community voices across languages — the default voice of Home Assistant's Assist. Status: the original rhasspy/piper repo is ARCHIVED (Aug 2025); active development moved to OHF-Voice/piper1-gpl (5.8k stars, active Oct 2026). Note the license change: the original was MIT, the maintained fork is GPL-3.0 — copyleft obligations apply to the version you should actually use. For new projects, go straight to piper1-gpl.
What it can do
- **Fully offline**: no cloud, no API keys, private by design
- **Fast**: real-time synthesis on modest CPUs (even Raspberry Pi)
- **100+ voices**: community voice library across languages
- **ONNX models**: small, portable voice files
- **Home Assistant**: the default Assist TTS engine
Setup
```bash
# Use the active fork:
git clone https://github.com/OHF-Voice/piper1-gpl
# Download a voice from the fork's releases, then:
# echo "Hello" | piper --model en_US-amy-medium.onnx --output_file hello.wav
```
Voices are small ONNX files (~60MB). No GPU or network needed.
Prerequisites
Any modern CPU; no GPU or API key needed. A voice model file (ONNX) from the releases.
Use cases
Works with
Trust score — 6/10
Scored 6/10 (metrics tier, metrics-v1). Evidence: 11,298 GitHub stars (log-scaled vs catalog peers); repository is archived — no future fixes expected. Manual review: defined local neural TTS (Home Assistant's voice), 11.3k stars; original repo ARCHIVED — use the active fork. License: Original: MIT (ARCHIVED Aug 2025); active fork OHF-Voice/piper1-gpl is GPL-3.0 — copyleft applies to the maintained version.
★ 11,298 GitHub stars
How the trust score is built
Full methodology →- Adoption · stars/downloads, log-scaled vs peers
- 7.7/10
- Maintenance · recency, releases, issue load
- 2.0/10
- Openness · license + docs
- 9.3/10
- Manual review · sanity read + red-flag check
- 7.0/10
- · 11,298 GitHub stars (log-scaled vs catalog peers)
- · repository is archived — no future fixes expected
- · manual review: defined local neural TTS (Home Assistant's voice), 11.3k stars; original repo ARCHIVED — use the active fork
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.