Developers building offline voice assistants, embedded devices and accessibility tools.

A fast, local neural text-to-speech engine optimized for Raspberry Pi and edge devices using ONNX models.

Self-hosted, runs fully offline on CPU, from embedded hardware up to servers.

Released around 2023 by the Rhasspy project, archived by its original maintainer in 2025 though still widely used.

Popular for privacy-preserving, low-latency speech synthesis without any cloud dependency or GPU requirement.

Uses ONNX Runtime and espeak-ng phonemization, no internet connection required at runtime.

Open source under MIT, self-hosted, no credentials, original repo archived in 2025 but over 11k stars and actively used via community forks.