Turn every spoken word into clean, usable text
Voice rewrites productivity.声音，重写生产力。TakeVox is a local-first speech-to-text and meeting assistant: an offline dictation engine, a live floating transcript, AI polishing and translation, and speaker-separated meeting minutes — with your audio staying on your device by default.
Free · no sign-up · data stays on your device
Latest release: —
Why TakeVox?
Four everyday pain points — one local-first answer.
How it works
From hotkey press to inserted text in under three seconds
Press & speak
In any app, press the global hotkey to start recording.
Live transcript
Watch the recognition appear in the overlay as you speak.
Stop to insert
Press the hotkey again — the final text lands at your cursor. ESC cancels.
Features
From dictation to meetings — the full path from speech to productivity.
Stop typing. Start thriving.不再打字，即刻迈向高效。
Meeting transcripts & minutes
Record or import audio for speaker separation, auto-generated minutes and action items with Markdown export.
Clipboard history
Automatically captures copied content with favorites, full-text search and instant paste.
Offline live dictation
A built-in local speech engine turns voice into text in real time, no network required.
Multilingual recognition & translation
Auto-detects Chinese, English and more; transcripts can be translated in full.
Global hotkeys
Start/stop dictation in any app with one shortcut; text is inserted at the cursor.
Floating live transcript
A translucent overlay shows transcription as you speak, with cool/warm themes.
AI polish & translation
Polish wording, rewrite style, or translate into many languages with a local or cloud LLM.
Professional glossary
Import industry term lists (CSV/JSON): misreadings are auto-corrected, domain terms are recognized right on the first pass, and cloud transcription syncs the same hotwords.
Local-first privacy
Audio and transcripts stay on your device; cloud transcription uses your own subscription key directly.
Who uses TakeVox?
Writing, meetings, study, development — anyone who turns speech into text.
Writing & interviews
Journalists, authors, creators: interviews become drafts directly, ideas captured instantly.
Meetings & teamwork
Real-time transcription, auto minutes and action items at a glance.
Study & lectures
Students and lifelong learners: transcribe lectures live, translate on the fly.
Development & notes
Developers: dictate design docs; clipboard history recalls code snippets.
Three modes, pick what fits
From “most private” to “zero setup”, TakeVox is the same app — only how your speech is processed differs.
Speak. Instant Produced.开口即成文。
Local mode
- Fully offline — audio and transcripts never leave your device
- Works without an account or network
- Runs entirely on-device after the first model download
Own API mode
- Bring your own LLM / transcription service
- OpenAI-compatible endpoint, free choice of model & parameters
- Share one service setup across your team
Subscription mode
- No model management — cloud transcription instantly
- Extended audio and speaker separation
- Synced across devices, pick up where you left off
Local-first vs. cloud transcription
Same speech-to-text — different data ownership, cost and availability.
| Dimension | TakeVox local mode | Traditional cloud STT |
|---|---|---|
| Data ownership | Audio & transcripts stay on device | Uploaded to provider servers |
| Offline use | Fully offline, no network needed | Requires a connection |
| Pricing | Free locally, unlimited | Per-minute / per-usage |
| Privacy | Stays on device by default | Depends on provider policy |
| Languages | ZH / EN / JA / KO / YUE | Varies by provider |
| Security | Built-in protection, encrypted by default | Varies by provider |
Enterprise plan
Deploy speech-to-text inside your organization: data stays internal, terminology matches your business, and privacy stays under your control.
On-premises deployment
- Deploy on intranet or private cloud — data never leaves your boundary
- Integrate with existing OA, meeting and document systems
- Unified accounts and role-based access per department
Terminology optimization
- Industry terms and proper nouns recognized first for accurate dictation
- Share a team-wide glossary for consistent output
- Keep improving as your business evolves
Privacy & internalization
- Audio and transcripts stay fully local by default
- Fine-grained access control and audit trails
- Meet internal compliance and data-retention policies
Download TakeVox
macOS supports Apple Silicon (ARM) only — no Intel; pick the matching Windows or Linux architecture below.
macOSARM only
Windows
Linux
Pro benefits
Subscribe in-app to unlock cloud transcription and meeting enhancements
FAQ
Which operating systems are supported?
How large is the offline model?
Is my audio uploaded to a server?
How does cloud transcription work?
Do you support professional glossaries?
Where are transcripts and clipboard data stored?
How do I pay for a subscription?
Can I export meeting minutes?
How do I get updates?
Ready to turn speech into text?
Free to download. Local-first, private by default.
Sign in to manage your subscription, usage and account