Short answer: Descript is a media editor built around the transcript of a recording you already made. If what you actually want is text appearing wherever your cursor is, the right alternative is a dictation tool. Voice Keyboard Pro puts your words into any Mac app or iPhone app as you work.
Most people who search for a Descript alternative are not shopping for a different video editor. They are trying to solve a problem one layer up: they want to write by talking, they tried the tool everyone recommended, and they discovered it is a production suite that expects a finished recording as its starting point.
That mismatch is worth naming clearly, because it decides which alternative is actually right for you. There are two different jobs hiding under the phrase "turn my voice into text," and the tools that do them well look nothing alike.
The archive job and the input job
The archive job is: something was said, and you want a record of it. A podcast episode. A recorded interview. A webinar. A video you need to cut down. The audio already exists as a file, the file is the source of truth, and the transcript is a working surface for editing or searching it. This is the job Descript is built around, and it is genuinely good at it. Editing media by editing text is a real idea that changed how a lot of people cut podcasts and videos.
The input job is: nothing exists yet, and text needs to appear. A reply in your inbox. A ticket note. A Slack message. A paragraph in a document. A comment on a pull request. There is no file, no recording, no timeline. There is a cursor blinking in a text field and a sentence in your head.
Almost everybody who writes for a living spends far more of the day on the input job than the archive job. You might record two hours a week. You almost certainly type for many more hours than that. A tool that only helps with the recorded hours leaves the bigger pile untouched.
A transcript of something you recorded is not the same as the email you still have to write.
Why a media editor feels like the wrong shape for daily writing
Say you want to answer a customer's question by talking instead of typing. In a media-editor workflow, the shape of that is: open the app, start a recording, say the thing, wait for the transcript, clean it up in the editor, select it, copy it, switch back to your inbox, paste it, fix the formatting, and send. Nine steps, an app switch in the middle, and a copy-paste at the end.
In a dictation workflow, the shape is: put your cursor in the reply box, hold a key, say the thing, release. The text is in the reply box. There was no file, no project, no export.
The step count is the entire argument. A workflow with nine steps gets used for the important stuff and abandoned for everything else, which is where most of your typing volume lives. A workflow with one step gets used for a two-line Slack reply, and that is what changes your day.
The friction nobody warns you about
The other thing that surprises people is the leftover work. A raw transcript is not finished writing. It has your false starts, your restarts, the sentence you abandoned halfway through. In a media tool that is fine, because the transcript exists to serve the recording. When the transcript is meant to be the deliverable, every one of those artifacts is homework you now have to do in a text editor.
A dictation tool that inserts one clean, complete utterance at your cursor skips that step. You speak a sentence, a sentence appears, you move on.
What Voice Keyboard Pro does instead
Voice Keyboard Pro is a voice-to-text app for Mac and iPhone. It is not a media editor and does not pretend to be one. It handles the input job.
On the Mac it lives in the menu bar. Hold a hotkey, speak, release, and the text appears at your cursor in whatever application is in front of you. That is the whole interaction, and it works system-wide, which means it works in your browser, your inbox, your editor, your CRM, your terminal, and every internal web tool your company has that will never have a dictate button of its own.
On the iPhone it is a keyboard with a microphone built into it. Switch to it with the globe key and you can dictate into any iOS app, including the ones that have gone out of their way not to offer dictation themselves.
Three features are worth calling out because they map onto complaints people have about transcript-first tools:
- Smart Vocabulary. A personal dictionary with replacement rules. Client names, product names, internal acronyms, the surname nobody's transcription engine has ever spelled correctly. This is the error class that no microphone upgrade and no amount of careful speaking will fix, because it is a vocabulary problem, not an audio problem.
- Meeting Mode. Speaker detection and AI notes on the Mac, with calendar meeting detection so it knows when a meeting is starting. No bot joins the call and appears in the attendee list.
- Voice Edit and translation on iPhone. Speak a change to fix text you already have, instead of hunting for the cursor with your thumb. Two-way translation while dictating across 24 languages, which matters if you write to people who do not read your first language.
The privacy difference is structural, not marketing
This is the part worth thinking about hardest, because it is not a matter of which company is more trustworthy. It is a matter of what each product has to do in order to work.
A media editing tool has to hold your media. Storing the recording is not an optional feature it could switch off; it is the product. Your files live in the tool, and everything follows from that, including how long they are kept and who at the company can reach them.
Voice Keyboard Pro is built for the input job, so it does not need an archive. Our server stores only operational pings. No audio and no transcript content. There is no library of your dictation to browse, because there is nothing to browse. Your history stays on your own machine.
We would still tell you to read our policy rather than take a blog post's word for it, and to read the policy of whatever you are comparing us against. If you want the general method for evaluating any dictation tool's privacy claims, we wrote one up in does voice to text store your audio, including the exact terms to search for in a policy document.
Honest cases where you should keep Descript
We would rather you keep the right tool than switch to ours and be annoyed. Keep the media editor if any of these describe you:
- You edit video or podcasts. Cutting a recording by cutting its transcript is a genuinely good workflow and we do not do it at all. This is not a close call.
- You have existing audio or video files to transcribe. Voice Keyboard Pro does not transcribe uploaded recordings. It transcribes you, live, into the field your cursor is in. If your input is a file, we are not the tool.
- You need multitrack, screen recording, or publishing. Production features are production features. We have none of them.
- Your team shares a media library. If several people need to search and reuse the same recordings, you want something built to store them.
- You need a verbatim record for compliance or legal reasons. Dictation gives you the text you meant to write, which is the opposite of a verbatim archive of what was said in a room.
None of that is a small list. If two or more of those lines describe your week, the honest recommendation is not to switch.
The middle path most people land on
The two categories are not really competing for the same slot, which means using both is a reasonable outcome rather than a cop-out. A podcaster we would expect to keep a media editor for producing episodes and add a dictation tool for the show notes, the sponsor emails, the guest outreach, and the ninety other small text jobs that surround one recording. We wrote about that specific pile in dictation for podcast show notes.
What usually changes is the mental accounting. Once dictation is available everywhere, people stop opening a recording tool for things that were never really recordings. The note you were going to record and transcribe becomes a note you just speak into the document.
If your real problem is meetings
Some people arrive at a media editor because they want meeting notes and it was the transcription tool they had heard of. That is a different problem again, and it is the one Meeting Mode is for: notes with speaker detection, no bot in the participant list, and calendar detection so it starts when the meeting does.
We have written direct comparisons for that category if it is where you actually are. The bot-free meeting notepad comparison covers the etiquette and consent side, and the Notta comparison goes deeper on why a searchable transcript library and finished text are two different deliverables.
How to test both in a week without wasting your time
Do not evaluate this by reading feature lists. Run it against your own writing for five working days.
- Day one: count. Keep a rough tally of writing tasks. Emails, messages, tickets, comments, docs. Do not change anything yet. Most people are surprised by how heavily the count skews toward short text in a browser.
- Day two: sort. Mark each one A if it started from a recording, B if it started from nothing. The A pile is the media editor's job. The B pile is dictation's job. The ratio tells you which tool your day is actually shaped by.
- Day three: install and set your hotkey. Pick a key you can hold comfortably and that does not collide with a shortcut you already use in your busiest app.
- Day four: load your vocabulary. Add the ten proper nouns you type most often to Smart Vocabulary before you judge accuracy. Accuracy complaints about any dictation tool are usually complaints about names, and this is the fix.
- Day five: dictate the B pile only. Do not force it on code, spreadsheet formulas, or anything with a lot of symbols. Type those. Speak the sentences.
At the end of the week you will have a real answer instead of a vibe. If your day turned out to be mostly A, keep what you have. If it was mostly B, you were never looking for a video editor.
What it costs
There is a free tier with daily limits, which is enough to run the test above. Pro is $4.99 a month or $34.99 a year, and includes both the Mac app and the iPhone keyboard.
Frequently asked questions
Can Voice Keyboard Pro transcribe a recording I already have?
No. It works live, at your cursor. If you have a file to transcribe, keep a tool built for files. We would rather say that plainly than let you find out after installing.
Does it work in the browser?
Yes, and this is usually the surprise. Because text is inserted at the cursor at the system level, it works in web apps without an extension and without anything being installed in a company tenant. That is often what makes it usable at work, where getting an integration approved is a project and downloading a Mac app is not.
Does it edit what I said?
Dictation preserves your words. It handles punctuation, capitalization, and filler; it does not rewrite your sentences into something you did not say.
Is it faster than typing?
For prose, comfortably. Most adults type around 40 words per minute and professional typists land somewhere around 80 to 100. Conversational speech runs 130 to 150 words per minute. That gap is why the input job is worth attacking at all, and it does not apply to code or spreadsheet work, where the keyboard stays the right tool.
What about my accent?
Modern transcription handles a wide range of accents well. When people report bad accuracy, it is far more often a proper-noun problem than an accent problem: the same five names wrong every time, rather than errors scattered evenly through the text. If your errors cluster on names, Smart Vocabulary is the fix.
The short version
If you record things and need to cut them, a media editor is the right tool and there is no reason to leave it. If you spend your day putting sentences into boxes on a screen, you were never looking for a media editor. You were looking for a faster way to type, and that is a different product.
Voice Keyboard Pro has a free tier. Install it, put your cursor in the next email you were about to type, hold the key, and say it. If that feels better than opening a project, you have your answer.