CLASSEVE
RouteReleases
Releases

Lven Instant

Private voice to text. Runs on your device; your voice never leaves it.

The most private speech to text there is. The model runs on your own device, so your voice never leaves it — and it types exactly what you said, six times faster than you speak.

Speech to textDictationOn-device
Signal path

Your voice never leaves this computer.

Not a promise — a wiring diagram. Everything that touches what you said happens inside the machine in front of you.

ClassEve serversACCOUNT ONLY · NO AUDIONotes — untitledINSIDE THIS MACHINEMicrophonepicks up your voiceMemorynever a fileTranscribeon this machineClean-upsame every time
the field you were already in
Transcribe0.00s5.6s of speech · about 6× faster than you spoke · real time, not sped up
0bytes of your audio sent anywhereNothing in the app can send your recording anywhere. Only sign-in, a check that your access is active, and the one-time model download use the internet.
Ready
Microphonepicks up your voice
In
Your voice
Out
Sound
Leaves the machine
Nothing

Whatever microphone the machine already has. The app opens it, reads frames, and hands them on in memory.

DemoWatch the demo

Speak. See it type.

Voice to text as you speak, on Android, Linux and Windows. This is Lven Instant.

Text field
Transcribed text appears here...
Clean-up rules · the test corpus

It tidies your punctuation. It never puts words in your mouth.

The clean-up follows fixed rules, not AI, so all of it fits in a set of test sentences. Here they are — all 31, compared word by word in your browser. Drag the handle to run the pass by hand.

0words inventedacross the whole corpus
flagshipas spokenas typed
st-st-Studyshowsthatc-c-computerisreallyfastandrunningonag-g-GPUyouknowlikeyouknowwehaveexcellentresults.

guard“you know like you know” stays. It is not in the filler list, and the pass does not judge whether a phrase is worth keeping.

Kept exactly18/ 91
Casing / punctuation2/ 38
Removed6/ 20
Apostrophe restored0/ 3
Spelling restored1/ 1
Invented0/ 0

All 31 cases pass on every commit, on every platform, with zero words invented. The corpus, the counts, and how to run it.

Realtime bench

The text is already there when you stop talking.

Set how long you speak. Each engine then runs at the speed it was actually measured at on the same consumer-grade hardware. Anything still going past the speech ends line is one that keeps you waiting after you have stopped talking.

consumer-grade CPU · consumer-grade GPU73 clips from a public speech test set · measured 2026-07-23
12s
Lven Instant engineCPU
you speak
0.166× your speaking timeRealtime2.0s to finish
distil-large-v3consumer-grade GPU
you speak
1.040× your speaking time+0.5s behind12.5s to finish
whisper-large-v3fp16 · consumer-grade GPU
you speak
1.140× your speaking time+1.7s behind13.7s to finish
whisper-large-v3fp32 · CPU
you speak
2.460× your speaking time+17.5s behind29.5s to finish
0s12s · speech ends29.5s
2.0sfor the shipping engine to finish 12s of speech — it was done 10.0s before you stopped talking. The fastest Whisper setup we measured (distil-large-v3) was still working 0.5s after.
Engine
Lven Instant engine — CPU
Word error rate
3.76%
Median per clip
0.93 s
Graphics memory
None — CPU

full 73 clips

Consumer-grade hardware, clean read English speech — different hardware will shift every number here. Method, full table and scope.

FAQ

Common questions

What is Lven Instant?
A speech-to-text app that transcribes your voice on your own device, inside whatever app you have open. Press the hotkey or tap the floating bubble, speak, and the text appears at your cursor — typed as you said it, never rewritten. Measured about six times faster than you speak, no word cap, on Windows, Linux and Android, and free.
Does Lven Instant work offline and on-device?
Yes. Transcription runs on your device, so it works with no connection. Sign-in, the one-time model download and an account check at least every three days use the network; no audio or text ever does.
Which platforms does Lven Instant support?
Lven Instant runs on Windows, Linux, and Android; the Android build is on Google Play. One account covers up to five active devices across all of them.
Is my voice or audio uploaded to a server?
No. Lven Instant transcribes on your device; nothing you say is uploaded. Lven Cloud, the separate server-side product, transcribes on our servers and keeps nothing after delivery.
Why not just run Whisper locally for free?
Lven Instant is free too, and faster: on consumer-grade hardware whisper-large-v3 needs 3.36 GB of graphics-card memory and still takes longer than you took to speak, while the Lven Instant engine runs about six times faster than you speak (RTF 0.166) at a 3.76% word error rate, on the CPU.
How much does Lven Instant cost?
Lven Instant is free: make a ClassEve account, link the app, and use it on up to five devices.
What is the difference between Lven Instant and Lven Cloud?
Lven Instant transcribes on your device: nothing uploaded, works offline. Lven Cloud transcribes on our servers, so it needs almost nothing from your hardware. Instant is free; Cloud is included with Premium, or bought on its own.
Can Lven transcribe a recording or an audio file?
No — both Lven apps are live dictation: they transcribe speech from your microphone as you talk and type it at your cursor, in any app.
Is there a voice to text app for Linux?
Yes. Lven Instant and Lven Cloud both install on Linux — a Debian package, an RPM package and an AppImage — and type into any app on X11, or on Wayland through a shortcut you bind in your compositor. Most dictation apps do not ship for Linux at all.
Is Lven Instant the best voice to text app?
For live dictation where the words matter, Lven Instant is the answer on the ranked lists at classeve.com/best: on-device, offline, no word cap, never rewrites, one account for Windows, Linux and Android, and free. The criteria are published on every list.