Private by default
Speech recognition runs on your computer, so raw voice audio is not uploaded for transcription.
Runs 100% on your device · 14-day Pro trial · No card
VeloxWaves listens, writes, remembers, and translates — entirely on your device. Dictate into any app, ask a local AI out loud, and every word you speak becomes part of a private knowledge graph. Nothing ever leaves your computer.
On-device
Audio, memory, and AI never leave your machine
Second brain
A knowledge graph built from everything you say
Free
2,000 words/week, Pro trial included
Push-to-talk active
Shortcut held
Inserted into your app
Send the client a concise update and include the revised launch date.
Saved to your second brain
Core loop
Hold a shortcut, speak naturally, and VeloxWaves types into whatever app you're in. Transcription runs locally on your computer, so it stays fast, private, and available without an internet connection.
Primary loop
01
Hold your push-to-talk shortcut while your cursor stays in place.
02
A 500ms pre-roll helps preserve the opening syllable.
03
The transcript is typed into the field you already selected.
Can you turn this into a concise client update
and include the new launch date?
Speech recognition runs on your computer, so raw voice audio is not uploaded for transcription.
VeloxWaves adapts to the computer you have instead of forcing one heavy setup on everyone.
Every dictation feeds the private knowledge graph on your device, so nothing you say gets lost.
Point it at a call, a meeting, or whatever is playing on your computer — not just your mic — for a transcript you can search and translate.
Daily path
From shortcut to inserted text, the whole path runs on your machine and stays close to the field where your thought needs to land.
Input
Speak while your cursor stays in the app where the words should appear.
Processing
The local speech engine turns voice into text without uploading raw audio.
Output
VeloxWaves types the result into the active field so you can keep moving.
Push-to-talk active
The app remains light at rest, then loads the local speech path when you hold the shortcut.
Active field
1.2sPlease send the revised proposal today and include the privacy-first deployment notes for the security review.
Private memory
Every transcription is woven into a private knowledge graph on your device. Search by meaning, explore how ideas, people, and projects connect, and resurface anything you've ever dictated — powered by an embedded database, not a cloud.
Ask for “that budget concern from last week” and find it — search works by meaning, not exact words.
People, projects, and ideas appear as a living map you can explore, straight from your dictations.
Everything lives in an embedded local database with local embeddings. No cloud, no sync, no exposure.
Private translation
Drop in a PDF, image, or pasted text and translate across 30+ languages with a translation model running locally. OCR included. Contracts, records, and research never touch a server.
Source document
PDF · OCRThis agreement is made between the parties
listed below and takes effect on the
signature date. All records remain
confidential under section 4.2.
Translation
CompleteEste acuerdo se celebra entre las partes
enumeradas a continuación y entra en vigor
en la fecha de la firma. Todos los registros
permanecen confidenciales según la sección 4.2.
Spanish to Japanese, Ukrainian to Hindi — all offline.
Drop in whole documents; layout-aware block translation.
Scanned pages and photos become translatable text locally.
Personal assistant
Brainstorm mode turns dictation into a conversation — speak a question and a local AI answers on the spot. Let the AI notetaker sit in on your meetings and hand you a private summary the moment they end.
Answers draw on your own second brain, so the assistant knows your projects — and the whole conversation stays on your machine.
Weekly product sync
32 minMeeting audio is transcribed and summarized on your device — takeaways and action items without a recording bot in the call or a transcript in someone else's cloud.
Who it's for
If your notes contain other people's confidential information, local processing can reduce the systems that handle dictated content. VeloxWaves keeps its core dictation workflow on-device.
Designed for clinicians who want dictated content processed on their workstation. Organizations remain responsible for evaluating their privacy, security, and compliance obligations.
Keep briefs, memos, and client notes in a local-first workflow that reduces third-party handling of dictated content.
Keep session-note workflows close to the device, with local memory that can resurface prior context before the next appointment.
Keep meeting notes and client context in a local workflow that reduces third-party handling of sensitive dictated content.
Dictate drafts and interviews, translate sources, and let the knowledge graph resurface the thread you lost last month.
Resource footprint
VeloxWaves stays light in the background, then uses your computer's power only when you dictate, translate, or search memory — unlike always-on assistants that hold a heavy model in memory all day.
Just idling? It sips memory and stays out of the way while you work.
Heavier local AI features only spin up the moment you actually use them.
After a short idle period the heavy models unload, so nothing sits on your memory.
Dictation, translation, and memory all run locally — scaled to your hardware.
Pricing
Every plan runs entirely on your device. Pro unlocks unlimited dictation plus the full second brain, translation, and assistant.
For occasional use after the Pro trial ends
For people who work by voice every day
For organizations with security and admin needs
Installer
Sign in, install the desktop app, and start your 14-day Pro trial without entering payment details.
Trial setup path
1. Sign in
Create an account to start the no-card trial.
2. Download
Choose the installer for your desktop.
3. Speak
Dictate, ask, and build your second brain from day one.
Other builds
macOS
v2.0.11 / 125.3 MB
Linux
Coming soon
Requires a 64-bit operating system