IMPORTANT: After installing the extension, open Preferences to download the required transcribe.cpp binary and an ASR model before using speech recognition. Turn your voice into text instantly, right on your desktop—no cloud, no accounts, no waiting. Plane-ASR is a GNOME Shell extension that transcribes speech to text in real time. Press a keyboard shortcut, speak, and watch the text appear directly at your cursor or be copied to your clipboard. Features • Start or stop recording with a single click from the top panel or by using a keyboard shortcut. • Live chunked transcription for long recordings, so text starts appearing while you're still speaking. • Transcribe existing audio files in addition to live microphone input. • Browse, download, verify, and select GGUF ASR models from Hugging Face directly in Preferences. • Use the provided CPU binary or your own Vulkan/CUDA build of transcribe.cpp. • 100% local processing through transcribe-cli (transcribe.cpp). No audio is ever sent to the cloud. Default keyboard shortcuts Toggle Recording Ctrl+Alt+Space Press once to start recording. Press again to stop. Push-to-Talk Ctrl+Shift+Space Hold to record. Release to stop.
Note: Binary files aren't shown on the web site. To see all files, please download the extension zipfile.
EGO-A-005 manual_review
extensions should not access the clipboard directly
Direct clipboard access via `St.Clipboard.get_default()` requires reviewer scrutiny.
Plane ASR
src/extension/indicator.js:267
St.Clipboard.get_default()
src/extension/indicator.js:301
St.Clipboard.get_default()
src/util/paste.js:38
St.Clipboard.get_default()
src/util/paste.js:74
St.Clipboard.get_default()
All is ok
Is this code generated by AI? If so, we have a rule for that: [EGO Review Guidelines: AI](https://gjs.guide/extensions/review-guidelines/review-guidelines.html#extensions-must-not-be-ai-generated)
The extension was developed with the assistance of AI tools as a productivity aid, primarily for consulting documentation, generating example implementations, and accelerating parts of the development process. All generated code was carefully reviewed, understood, tested, and adapted before being incorporated into the project. The extension has been thoroughly tested on multiple systems and hardware configurations to verify its functionality, stability, and performance under real-world usage. Every feature has been validated to ensure it behaves as intended and meets the project's quality standards.
I compiled the transcribe.cpp binary myself with a focus on broad CPU compatibility, allowing the extension to run on a wide range of systems without requiring users to build the project themselves. During testing, I verified that it works not only on modern hardware but also on an approximately 10-year-old ThinkPad, confirming that the CPU backend performs reliably across different generations of hardware. While evaluating existing speech-to-text solutions, I found that many offered similar functionality but were difficult to integrate into a GNOME workflow or required additional tools and libraries to automatically insert the transcribed text into the active application. To provide a simpler, self-contained, and more user-friendly experience, I decided to develop this GNOME extension.
ok