Maise
On-Device Speech for Android
A free, open-source Android speech engine with on-device TTS and ASR. 40+ voices, 9 languages, Whisper transcription — zero internet required.
Features
Fully On-Device
Speech synthesis and recognition run entirely on your Android device using ONNX Runtime. Your audio and text never leave your phone.
40+ Voices
Powered by the Kokoro engine with voices across English (US/UK), German, French, Greek, Italian, Japanese, Portuguese, and Chinese.
Whisper ASR
On-device automatic speech recognition via Whisper, enabling accurate offline transcription without any cloud dependency.
System-Wide TTS
Works as a native Android TextToSpeech service — any app that uses the standard TTS API can use Maise automatically.
Streaming Playback
Producer-consumer architecture enables low-latency streaming audio output as synthesis happens, rather than waiting for the full result.
Voice Preview
The companion app lets you browse voices, preview synthesis in real-time, and set your system-wide default.
Maid Integration
Pairs with Maid for fully on-device AI conversations — local inference, local voice, and local transcription, end to end.
Supported Languages
- English (US)
- English (UK)
- German
- French
- Greek
- Italian
- Japanese
- Portuguese
- Chinese
Guides
Step-by-step guides for setting up and getting the most from Maise.
Set Up System TTS
Make Maise the default voice engine for your entire Android device in a few taps.
Read guide →Set Up Speech Recognition
Enable offline voice dictation in any app using Whisper-powered on-device ASR.
Read guide →Use Maise with Maid
Combine local TTS and ASR with Maid's on-device AI for a fully offline assistant.
Read guide →Voices & Languages
Browse all 68 Kokoro voices across 9 languages and learn how to preview them.
Read guide →Maise is MIT licensed and actively developed. v1.0.0 released February 2026.
Contribute on GitHub →