Fully Offline, One-Time Purchase
AI Speech Recognition & Transcription
Cloud-based speech recognition is convenient, but sending sensitive data to external servers is a concern...
Need voice input on air-gapped environments or development servers with network restrictions...
Writing documentation while working in the terminal. If only you could just speak and type...
Plenty of options for Windows/Mac, but hard to find a quality speech recognition app for Linux...
Runs OpenAI Whisper locally.
Your voice data never leaves your machine.
Safe to use even in air-gapped environments.
On machines with NVIDIA GPUs, CUDA hardware acceleration delivers incredible processing speed.
💡 Works on CPU too when no GPU is available (auto-detected)
*Compared to CPU with large-v3-turbo model
Works with local LLMs (Ollama, etc.).
Completely offline and free AI features.
No API key required. No cloud communication. Full privacy protection.
Simply give voice instructions on selected text.
"Translate to Japanese", "Summarize this", "Make it formal", etc.
Powered by local LLM — works offline
Just say "Search for ..." and it automatically opens a browser search. Research made effortless.
State-of-the-art speech recognition powered by OpenAI Whisper. Excellent for both Japanese and English.
Start recording with a single hotkey from any app. No need to interrupt your workflow.
Recognition results are automatically typed into the active window. No copy-paste needed.
Automatically apply different settings for each app. Separate configs for VSCode, Slack, and more.
Auto-format recognized text with local LLMs (Ollama, etc.). Grammar correction, formalization, and more. Completely offline and free.
Automatically fix common misrecognitions. Regular expressions supported.
Edit text with voice commands like "new line", "period", and more.
Say "Search for ..." to automatically open a browser search.
Launch apps by voice. Say "Open VS Code" and it starts.
Give voice instructions on selected text. Translation, summarization, and more via local AI. No API key required.
Register technical terms and proper nouns to improve recognition accuracy.
Audio cues for recording start/stop. Stay informed even when looking away.
Transcribe multiple audio and video files at once. Supports MP4, MKV, and more. SRT/VTT subtitle export available.
Register frequently used AI tasks to dedicated hotkeys. Translation, formal language, summarization — up to 20 custom instructions, each triggered with a single key.
All recognition results are automatically saved. Search, copy, and reuse anytime.
Floating indicator at the bottom of the screen during recording. Real-time audio level waveform display.
Easy-on-the-eyes dark mode built in. Comfortable for extended use.
sudo apt install pulseaudio-utils xdotool xclip ffmpeg
For Ubuntu/Debian. Use dnf for Fedora, pacman for Arch.
chmod +x RocketWhisper-*.AppImage
./RocketWhisper-*.AppImage
No installation required. Just download and run.
Press to talk. Press again to recognize. That's it.
| Ubuntu | 20.04 LTS or later |
| Debian | 11 (Bullseye) or later |
| Fedora | 35 or later |
| Linux Mint | 20 or later |
| Pop!_OS | 20.04 or later |
| Arch Linux | Rolling |
| ARM64 (aarch64) | ✅ Recommended (DGX Spark, etc.) |
| x86_64 (AMD64) | ✅ Supported |
pulseaudio-utils |
Microphone recording (parec) |
xdotool |
Keyboard automation |
xclip |
Clipboard access |
ffmpeg |
Audio conversion (optional) |
| DGX Spark | Blackwell ✅ Optimized |
| Jetson AGX Orin | Ampere ✅ |
| Jetson Orin NX/Nano | Ampere ✅ |
| Falls back to CPU when no GPU is present | |
| X11 | ✅ Fully supported (recommended) |
| Wayland | ⚠️ Limited support |
| Some features are limited under Wayland | |
| small/medium model | 8GB or more |
| large-v3-turbo | 8GB or more recommended |
| large-v3 | 16GB or more recommended |
The following features are limited under Wayland:
ydotool required, may need root privileges)wl-clipboard required)sudo apt install ydotool wl-clipboard
💡 Log in with an X11 session for full functionality
Running RocketWhisper requires installing dependency packages.
Please install the required system packages for microphone recording and keyboard automation beforehand.
sudo apt install pulseaudio-utils xdotool xclip ffmpeg
For detailed installation instructions, see Help: Installation.
Package lists and commands for each distribution are available at Help: Required Packages.
Write documentation while working in the terminal. Input text without leaving the keyboard.
Safe to use even in environments handling confidential data. No data is ever sent externally.
Works in air-gapped and network-restricted environments.
Leverage DGX Spark and Jetson hardware. Blazing-fast recognition with CUDA acceleration.
chmod +x. On systems without FUSE, you can extract it with --appimage-extract.