Speed and privacy, together.
Turn speech into 🔨 Creator work
On-device first. A quiet recording HUD stays out of the way, with FunASR / Whisper and audio/video transcription.
On-device first
Shortcut input
Audio + video
macOS 12+ Apple Silicon · Windows x64
00:08Recording
Keep your window. The text returns to the app in focus.Ousia Voice
Works with
Speak it, and the text returns to your work.
For messages, notes, and longer writing, press a shortcut and speak instead of switching tools or typing everything by hand.
Current app
Cursor
Issue draft: context, reproduction steps, current conclusion, next action.Turn speech into usable text.
Instant input and technical discussion transcripts in one desktop tool.
Voice typing
Press a shortcut, speak, and enter text
For prompts, bug reports, code comments, commit messages, and quick technical notes. Text returns to the current tool, reducing copy and paste.
File transcription
Turn audio and video into text first
For requirement reviews, debugging calls, technical meetings, and video material. Generate text or draft subtitles before review and editing.
meeting_audio.m4aDraft subtitles
84%Faithful
Keep your meaning
Ousia Voice aims to preserve what you said. When needed, it can add punctuation and apply your common replacements.
Polish
Optional text cleanup
If you want smoother wording, turn cleanup on. If not, Ousia Voice simply transcribes.
BeforeKeyboard
Think context · Type prompt · Fill gaps · Explain again
AfterOusia Voice
Shortcut · Speak the full context · Return to tool · Tweak
| Capability | Ousia Voice | Apple Dictation | Windows voice input | ChatGPT | Claude | Wispr Flow |
|---|---|---|---|---|---|---|
| Global shortcut | ||||||
| Returns to app | ||||||
| On-device first | ||||||
| File transcription | ||||||
| Custom dictionary | ||||||
| Optional AI cleanup |
Optimized for your focus.
From the first spoken thought to the final commit, Ousia Voice stays in your tool.
Your privacy comes first.
Local models
Installed local models are preferred by default; recordings and transcript history are not saved by default.
History is your choice
Recordings and transcript history are off by default. You can keep them locally when you want, and clear them anytime.
Online features are optional
If you choose online recognition or text cleanup, Ousia Voice makes that clear in settings before you use it.
What you may want to know about Ousia Voice first.
01What can Ousia Voice do?
It turns speech into text and sends it back to the current app. It can also turn recordings and videos into text or draft subtitles.
02Which recognition modes are supported?
FunASR is the default for Chinese. Whisper models are available for English and multilingual workflows, and are managed in the desktop app.
03Does it require an internet connection?
Local recognition and shortcut input do not require the internet by default. AI organization, high-accuracy recognition, and model downloads use the network when enabled.
04Which systems are supported?
Current installers target Apple Silicon Macs and Windows x64. macOS 12.0 or later is required.
Download Ousia Voice. Speak first, edit second.
Current version v1.0.0Updated 2026.07.08
