Features
From capturing your voice to delivering polished text — discover everything VoxCraft can do.
Capture
Meeting notes from today. We discussed the Q4 budget and timeline for launch.
Tap to toggle hands-free · hold for push-to-talk.
Start, stop, cancel — one click or one shortcut.
Meeting notes from today. We discussed the Q4 budget. Plus the product roadmap timeline.
Meeting notes from today. We discussed Q4 budget and the timeline.
✓Click any word to edit — live or after recording.
50+ languages — dictate in whichever you prefer.
The meeting is at 2pm actually 3pm. Bring cake.
The meeting is at 3pm. Bring cake.
No keyboard, no click — just speak.
Speak naturally and see your words appear on screen instantly. VoxCraft shows both a live speculative preview and finalized text, so you always know where you are.
Start recording, say “Meeting notes from today. We discussed Q4 budget and the timeline for the product launch.” — text appears word-by-word, then solidifies once you pause.
Record without touching the mouse. A quick tap of Ctrl+Shift+R toggles hands-free recording on and off. Hold the same shortcut for push-to-talk — release to stop. VoxCraft detects hold vs. tap automatically based on how long you keep the keys pressed.
Tap Ctrl+Shift+R to start, tap again to stop — or hold it down for quick bursts in the middle of a meeting.
Full control over every session. Start to begin, Stop to finalize, Cancel to discard instantly. Every control is one click away and mirrored on the keyboard shortcut.
Start a recording, speak for a few seconds, then hit Cancel — the transcript is discarded cleanly. Start again — Stop to keep the result.
Paused halfway through a thought? Hit Resume and VoxCraft picks up exactly where you left off, appending new speech to the existing transcript instead of starting fresh.
Dictate one sentence, stop, think, then click Resume and keep going — the new text flows into the same paragraph.
Click any word in the live transcript and fix it without breaking stride — mid-recording or after. If you edit after a recording ends, VoxCraft automatically regenerates the LLM response from your corrected text.
Dictate a sentence, click on any word to edit it inline, confirm with the green tick — then keep talking. Edit after stopping and the AI response refreshes automatically.
VoxCraft automatically formats numbers, times, currency, dates, lists, and punctuation as you speak. Say “one thousand” and get “1,000”. Say “three oh five p.m.” and get “3:05 PM”.
Say “five dollars” → “$5” · Say “January fifteenth” → formatted date · Say “comma” or “period” to insert punctuation.
Dictate in over 50 languages and regional variants — including English (US, UK, AU, IN, NZ), Spanish (incl. Latin America), French (incl. Canada), German (incl. Switzerland), Portuguese (Brazil & Portugal), Italian, Dutch, Flemish, Japanese, Korean, Hindi, Mandarin, Russian, Arabic, Turkish, Vietnamese, Thai, Hebrew, Polish, Swedish, and many more.
Switch to Spanish in Settings > Language, then dictate “Hola, me llamo Sarah” and the app transcribes in Spanish.
Correct yourself naturally while speaking. Say “actually” or “I mean” and the AI understands you’re revising what you just said — replacing the mistake with your correction automatically.
Say “The meeting is at 2 PM… actually 3 PM” — the AI knows you’re correcting yourself and outputs “The meeting is at 3 PM”, not both times.
Say “Hey Vox” and VoxCraft starts recording instantly — no keyboard, no clicks. On-device detection runs quietly in the background, so you can go straight from thought to dictation.
Enable Wake word in Settings, then say “Hey Vox” — recording starts automatically and you can dictate straight away.
Command
Had a call with the client. They like the mockups but the timeline is tight — budget is limited and they want to launch before September.
Output Hi Sarah, thanks for the call today. The mockups look great — just wanted to flag that the timeline feels tight given the budget…
Notes from the standup. Backend deploy is blocked on the API review. Sarah will follow up by Thursday. slash stop
I’ll see you on TuesdayWednesday.
Voice-activated backspace — removes the last word.
So the budget is tight but we can push the launch to March…
Throw the take away — start fresh.
Rename any voice keyword in Settings.
Meeting summary. slash commandMake into email.slash command Thanks.
Select any phrase — wrap it in a command retroactively.
Contact me at slash my email[email protected] any time.
Trigger phrases auto-expand into real text.
hello slash comma how are you slash question mark
Hello, how are you?
we need to ship this Friday or we’re toast
✓ Output We need to ship this by Friday, or we’ll miss our window.
Email new win about the project.
Email Nguyễn about the project.
Say “slash command” mid-dictation to open a command block. Everything inside becomes instructions for the AI — not dictated text. Say “slash command” again to close.
Insert command blocks anywhere in your dictation — at the beginning, middle, or end. Add as many as you like; all instructions are concatenated and sent to the AI together.
Say “slash stop” to stop recording and trigger AI processing instantly. Your dictation is sent to the LLM, the result renders on screen, and the output is auto-copied to your clipboard — ready to paste.
Dictate a few sentences about your day, then say “slash stop” — recording stops, the AI processes your text, and the result lands in your clipboard.
If a command block is open when you say “slash stop”, it auto-closes before processing — so you never need to remember to close it yourself.
Say “slash delete” to remove the word immediately before it — a voice-activated backspace. Perfect for fixing a quick mistake without breaking stride.
Say “I’ll see you on Tuesday slash delete Wednesday” — “Tuesday” disappears and “Wednesday” takes its place.
Changed your mind mid-sentence? Say “slash cancel” and VoxCraft stops recording and throws the transcript away — nothing hits your clipboard. Rename the trigger phrase to anything that feels natural.
Start dictating, realise you want a clean take, then say “slash cancel” — the transcript vanishes and you’re ready to start fresh.
Don’t like saying “slash command”? Rename it to anything you want — “hey AI”, “computer”, “note”. Same goes for slash stop and slash delete. All keywords are case-insensitive.
Change the keyword to “hey AI”, then dictate: “Draft a proposal hey AI make it persuasive hey AI” — works exactly the same.
Forgot to say “slash command” at the right moment? In Edit mode, click on any phrase in the transcript and wrap it in command keywords retroactively — the AI treats it as an instruction when the text is processed.
After dictating, enter Edit mode, click where you want the block to start and type “slash command” — then do the same at the end. The text between the keywords becomes an AI instruction.
Create trigger phrases that auto-expand into longer text. Default snippets include “slash my email”, “slash my name”, “slash my work email” — or create your own for signatures, addresses, URLs.
Say “Contact me at slash my email any time” — the trigger swaps instantly for your saved address.
Dictate punctuation with natural triggers like “slash comma”, “slash period”, “slash question mark”. Rename any trigger to whatever feels natural to you.
Say “hello slash comma how are you slash question mark” — produces “Hello, how are you?”. Prefer shorter triggers? Rename “slash comma” to just “comma” in Settings.
Advanced snippets wrap your dictation in an AI instruction automatically. Say “slash make professional” once and the snippet expands into a full command block — no need to speak the instruction yourself.
Say “we need to ship this Friday or we’re toast slash make professional slash stop” — AI rewrites it in a professional tone and auto-copies.
Add names, brands, acronyms, and domain jargon to improve transcription accuracy. These terms are sent to the speech engine as emphasis hints — so an unusual name like “Nguyễn” gets spelled correctly every time.
Add a colleague’s name, your product’s brand, or a medical term to your dictionary — transcription quality jumps immediately.
Transform
so the project timeline looks good um we should be done by friday and costs are under budget
So the project timeline looks good. We should be done by Friday and costs are under budget.
The project timeline is on track. We should be finished by Friday, and costs remain under budget.
Project update: the timeline is on track for a Friday delivery, and current spending remains under budget. No blockers at this time.
hey so I had a really good meeting with the client today um they loved the designs and want to move forward with phase two
Output Had a great meeting with the client today. They loved the designs and want to move forward with phase two.
had a call with Sarah about the project timeline and budget
Output Hi team, I spoke with Sarah regarding the project timeline and budget. Here’s a quick update…
Output Meeting with Sarah — Timeline: on track. Budget: under review. Action: Sarah to confirm by Thursday.
so the budget is tight but we can make it work if we cut the travel expenses and push the launch to March also the team is overstretched
Output The team is currently overstretched, and the budget is tight. To move forward, I recommend we cut travel expenses and push the launch to March.
Notes for the team on our branching strategy
Output Notes for the team on our branching strategy: Use main for releases, develop for integration, feature/* for new work, and hotfix/* for urgent fixes. Merge via PR.
the auth tokens are expiring before the session times out
Hey Jake, here is that bug I found
Output Hey Jake, here is that bug I found: Auth tokens expire before the session timeout, logging users out mid-task. Needs a token-refresh hook.
Output Heads up team — found a bug in auth: tokens expire before the session timeout, causing random logouts. Looking into a refresh hook. Will update with a fix.
Every dictation is sent to an AI for transformation. Choose how aggressive the AI is: Light fixes grammar and filler words. Balanced restructures for clarity. Thorough rewrites completely into polished prose.
Need more control? Switch the mode to Custom and write your own default instruction — or choose from built-in presets like “Rephrase for email” or “Format as markdown”.
The simplest workflow: just dictate and the AI automatically cleans up your speech — removing filler words, fixing grammar, and polishing the result. No commands or special keywords required.
The AI processes your dictation according to the active mode and instructions. You can change the mode or add command blocks for more control — but by default, it just works.
Switch the AI mode to Custom and choose a default instruction that runs every time. Pick from built-in presets like “Rephrase for email” or “Format as meeting notes” — or write your own from scratch.
Your default instruction runs automatically on every dictation. Switch presets any time to change how the AI transforms your text — no need to repeat yourself.
Embed instructions directly in your dictation using command blocks. Say slash command to open a block, speak your instruction, then slash command again to close. The AI treats everything inside as instructions — not dictated text.
Custom commands overwrite the default behaviour. However, the model will still reference your custom instructions and perform the action. It treats your instructions as skills you can combine: e.g. “make professional and reorder for logical progression of the argument”.
Commands aren’t just for reformatting — they can generate new content. Ask the AI to add a code reference, write a paragraph, insert boilerplate, or expand an idea. The generated content merges seamlessly into your output.
Say “Notes for the team on our branching strategy slash command add a summary of the Git Flow branching model slash command slash stop” — the AI writes the summary for you.
Tell the AI to save information as a note. Dictate a bug report, deploy steps, meeting action items — anything you want to remember. The AI stores it as a named note you can pull up later by voice.
Say “the auth tokens are expiring before the session times out slash command add this to my notes slash command slash stop” — the AI saves it as a note called “Auth Token Bug”.
Reference your saved notes mid-dictation. The AI reads the note content and weaves it into your output. Store bug reports, deploy steps, code snippets, reference material — then pull it into any conversation by voice.
First save a note with a bug report. Then say “Hey Jake, here is that bug I found slash command attach the auth token bug from my notes slash command slash stop” — the AI retrieves and inserts it.
You don’t need to dictate anything at all. Just open a command block, give the AI an instruction, close the block, and say slash stop. The AI generates the entire response from your instruction alone.
Say “slash command grab the auth bug note and rephrase it into a Slack message for my team slash command slash stop” — the AI retrieves the note and drafts the message.
A searchable popup that shows all your snippets, templates, and notes in one place. Filter by type, search by name or content, and click any item to copy it to your clipboard instantly. Access it with a keyboard shortcut from anywhere.
Press Shift+/ on web (or Ctrl+0 on desktop) to open the Prompt Library. Type to search, arrow keys to navigate, Enter to copy.
Deliver
# Q3 Plan **Goals:** - Ship beta - *Hire two PMs* > Budget locked
Met with SarahSara about the Q3 roadmap and budget review.
so um yeah I think the budget is like fine for now but we should probably revisit it next month maybe
The budget is sufficient for now, but we should revisit it next month.
Copy transcript text with keywords and command blocks automatically stripped — only clean dictated content goes to clipboard. AI responses are auto-copied the moment processing completes. On the desktop app, auto-paste inserts text into your active application - and you can take it one step further with Hands-Free Send.
Say "Hello world slash command improve tone slash command how are you" — clipboard gets "Hello world how are you", commands stripped.
Skip the keyboard entirely. Once auto-paste drops your AI response into a chat app, search box, or prompt, VoxCraft can press Enter for you - turning dictation into a true voice-only workflow. Toggle it on by default in Settings, or trigger it per-dictation with a voice keyword.
Two voice keywords let you override your default for a single response: say “slash send” to force auto-paste plus Enter, or “slash paste” to paste without sending - even if your defaults say otherwise. Both keywords are configurable in Settings.
Focus a chat app, dictate your message, and finish with “slash send” - VoxCraft processes the response, pastes it, and presses Enter. Your hands never leave your lap.
All transcriptions are saved and organized by date. Search across all transcripts and AI responses in real-time. Filter by date range for precise lookup. Toggle between viewing original dictation and AI-transformed output.
Search your history for "budget" to find all transcripts mentioning that term. Combine with date filtering for precise results.
Transcripts and AI responses render with full markdown formatting: bold, italic, headings, lists, code blocks, links, blockquotes, and horizontal rules.
Tap the edit button on any transcript to refine the text directly. Fix a misheard word, restructure a sentence, or clean up your train of thought — something you simply can’t do in other speech-to-text apps where you’re stuck with audio you can only reprocess.
After dictating meeting notes, notice a name was transcribed wrong? Tap edit, fix it in seconds, and your corrected version is saved and ready to copy.
Toggle between your raw dictation and the AI-processed version on any transcript. See exactly what words were changed, restructured, or polished — full transparency into what the AI did.
After an AI-enhanced dictation, click the toggle on the transcript to flip between your original spoken words and the polished output. Compare side by side to build confidence in the AI’s work.
Your Setup
Distinct audio feedback for every action: start, stop, cancel, chime, pop, delete, command open/close. All individually toggleable.
Settings > GeneralCustomizable shortcuts for key actions. Web: Ctrl+Shift+R to record, Shift+/ for prompt library. Desktop: global Ctrl+Shift+Space works from any app.
View monthly stats with daily bar charts: words transcribed, AI tokens used, audio duration, streak. Basic (free): 1,000 words/week. Pro ($8/mo): 100,000 words/week.
Settings > Plans & BillingToggle between dark (default) and light modes. Respects your system color scheme preference on first visit.
Settings > General > Dark ModeVoxCraft runs in any modern browser. Access your transcriptions across devices and platforms — all you need is an internet connection.
Native app for Windows, macOS, and Linux with global shortcuts, overlay window, system tray, and automatic updates.
All settings, notes, transcripts, and history sync across devices in real-time. Edit profile, manage subscription, or delete your account.
Workflows
Press Ctrl+Shift+R to start recording
Speak naturally: "Meeting notes — discussed Q4 budget, agreed on new timeline, action item for Sarah to prepare the report"
Press the shortcut again to stop
Clean text saved to history and clipboard
Record and speak: "hey so the customer called about the outage thing slash command rewrite as a professional incident summary slash stop"
AI transforms into a polished incident summary
Result auto-copied to clipboard, paste into Slack or email
Set "Rephrase for email" as your default instruction
Record: "hey John, wanted to follow up on our chat yesterday, I think we should go with option B because it saves us two weeks slash stop"
AI polishes into professional email body
Paste directly into your email compose window
Record: "We need to ship this feature by next Friday slash make professional slash stop"
The "slash make professional" snippet injects a command block automatically
AI rewrites with professional tone
Auto-copied, ready to paste
Morning: Record meeting notes, stop
Afternoon: Go to history, click Resume on the morning's transcript
Continue dictating — new text appends to the same item
Stop when done — full day's notes in one entry
Settings > Language > Select "French"
Record and speak in French
Transcription appears in French
AI can still process with English instructions if desired