On-device transcription
Speech recognition runs locally on Windows, Linux and Android. No microphone audio ever crosses the network, and once your entitlement is cached it transcribes with no connection at all.
On-device voice-to-text for Windows, Linux, and Android that types what you said, with the punctuation.
It does not rewrite you, and it does not upload your voice. Speech recognition runs on your own device; account and entitlement checks are separate requests that carry no audio.
Lven is a floating voice-to-text overlay that drops transcribed speech straight at your cursor in whatever app has focus. Tap the bubble or press the hotkey, speak, and the text appears.
It runs on Windows, Linux, and Android. macOS and iPhone are in internal testing with no public release date. On shipped platforms, microphone audio is processed by the local transcription engine rather than sent to ClassEve for remote inference.
The app downloads and verifies speech-model files during setup or repair. It also contacts the ClassEve API for profile and entitlement checks and refreshes account tokens before they expire. Those requests are separate from the dictated-audio path. Lven Cloud is the separate server-transcribed option.
Real-time voice transcription across Android, Linux, and Windows. This is the experience you get with Lven.
Paddle shows the exact trial end, first-charge date, and cancellation cutoff before checkout is confirmed.
Choose a planSpeech recognition runs locally on Windows, Linux and Android. No microphone audio ever crosses the network, and once your entitlement is cached it transcribes with no connection at all.
A deterministic pass clears stutters and fillers, restores contractions and proper nouns, and adds punctuation. No language model, no network — the same audio always gives the same text, never a rewrite.
After two months of membership, the app downloads an Ed25519-signed licence. Every build published while you paid keeps working once you cancel — offline, no server, no expiry.
A bubble sits at the screen edge with a global hotkey. Tap or hold, speak, release. Windows places text through UI Automation and Linux through AT-SPI2, each falling back to the clipboard.
Lven Instant currently publishes builds for Windows, Linux, and Android. macOS and iPhone builds are in internal testing and have no public release date.
The product registry records an exact version, byte size, download URL, and SHA-256 for every available platform. Planned platforms have no artifact or release date.
Lven Instant performs speech recognition on the device, so its transcription path does not upload microphone audio for remote inference.
Actual completion time depends on the device, model state, audio length, and operating conditions. ClassEve does not publish one universal latency number for all hardware.
Cloud transcription sends audio through remote infrastructure and depends on a working network path. Local transcription reduces that exposure for dictated audio and removes the remote-inference stage, while shifting processing requirements to the device.
Lven Instant narrows the trade. Its on-device model is optimized for responsive everyday dictation: no speech-audio upload and no inference round trip. Because microphone audio and transcript text are not sent to ClassEve for transcription, they are not present in our transcription infrastructure.
Lven Cloud is a separate server-transcribed path for users who prefer cloud processing or whose hardware does not meet the on-device requirements. The two products have different data paths and separate entitlements.