Releases: suinswofi/lectrote
Release list
Lectrote 1.5.6 + speech (unofficial fork build)
Unofficial build of Lectrote 1.5.6 with added speech output and voice input.
This is a fork build, not an official Lectrote release. The speech features have been offered to the main project as erkyrath/lectrote#185; until they land there, these builds let you try them today. Everything else works exactly like official Lectrote 1.5.6.
What's added
Both features run entirely on your computer — no cloud services, and nothing you say or play is sent anywhere. Turn them on in Preferences (Audio tab) or the View menu.
- Speak Story Text (Ctrl-Shift-V): reads each turn's new story text aloud with the Kokoro neural voice — pick from several voices and set the speaking rate. Stop Speaking is Ctrl-
., Repeat Last Turn is Ctrl-Shift-R. - Voice Input (Ctrl-Shift-M): push-to-talk — hold the talk key (Ctrl-Space by default, configurable, including bare Ctrl/Alt or a custom key), say your command, release. It's recognized locally with Whisper and typed into the game (sent immediately, or left for you to edit — your choice). Listen (Toggle) is Ctrl-Shift-L.
The first time you enable a feature, Lectrote downloads the model it needs from huggingface.co into your Lectrote data folder (~90 MB for the voice; 130–900 MB for the recognizer depending on the size you pick). After that it works fully offline.
Downloads
| Platform | File |
|---|---|
| Linux (x64) | Lectrote-1.5.6-speech.1-linux-x64.zip |
| Linux (ARM64) | Lectrote-1.5.6-speech.1-linux-arm64.zip |
| Windows (x64) installer | Lectrote-1.5.6-speech.1-WinInstall.exe |
| Windows (x64) | Lectrote-1.5.6-speech.1-win32-x64.zip |
| Windows (ARM64) | Lectrote-1.5.6-speech.1-win32-arm64.zip |
| Windows (32-bit) | Lectrote-1.5.6-speech.1-win32-ia32.zip — no speech features (there is no 32-bit ONNX runtime) |
Notes:
- Windows: the builds are not code-signed, so SmartScreen may warn you ("Windows protected your PC"). Choose More info → Run anyway if you trust this fork.
- Linux: depending on your configuration you may need to launch with
--no-sandbox. - Mac: no builds here — official Lectrote Mac releases are signed and notarized by the original author, which this fork cannot do. Mac users can build from source or use the official release until the feature arrives upstream.
Credits and license
Lectrote is by Andrew Plotkin (MIT license); this fork only adds the speech features. Speech uses kokoro-js, Transformers.js, and ONNX Runtime; the Kokoro model is Apache-2.0 and the Whisper models are MIT. See the repository LICENSE file for full details.