How to Dictate in Roam Research: Voice-to-Text for Blocks, Daily Notes, and Your Graph

Short answer: To dictate in Roam Research, install Voice Keyboard Pro on your Mac or iPhone. On Mac, click into any block, hold your hotkey, and speak; the text lands at the cursor in seconds. On iPhone, tap the mic on the Voice Keyboard Pro keyboard inside Roam in Safari or your browser. Add page links and tags with a tap afterward.

Roam Research is a tool for thinking, and thinking is faster out loud than it is through your fingers. The whole appeal of Roam is that you capture an idea the moment you have it and let the graph connect it to everything else later. But that appeal depends on capture being frictionless, and typing is the friction. You have the thought fully formed, then you slow down to a crawl getting it into a block. By the time it is written, half of what you meant to add has already slipped away.

Voice fixes the bottleneck at its source. The average adult types around 40 words per minute, while comfortable speech runs 130 to 150 words per minute, so speaking is roughly three times faster than typing for most people, and it takes far less deliberate effort. When getting a thought into your graph is as quick as saying it, you capture the ones you would otherwise have lost. This guide covers how to dictate in Roam Research on both Mac and iPhone, how to handle blocks and bullets by voice, how to add page links and tags cleanly, and how to build a daily-notes capture habit that actually sticks.

Why dictate in Roam Research?

Roam is unusual among note apps because it is built around small units of text. Every line is a block, every block can be referenced from anywhere, and the value of the whole system compounds as you add more of them. That structure makes Roam extraordinarily powerful and also slightly punishing, because it rewards volume. The people who get the most out of Roam are the ones who put the most into it, and the barrier to putting things in is how fast you can write.

Dictation removes that barrier. When you can speak a block instead of typing it, the cost of capturing a thought drops close to zero, and you start recording the small observations you used to skip. The tasks in Roam that benefit most from voice are the ones that are pure prose:

• Daily notes: the running log of what you read, thought, and did, which is the heart of most Roam workflows.
• Literature notes: your own summary of an article, book, or paper in your own words, which is exactly what voice is best at.
• Fleeting thoughts: the half-formed idea you want out of your head before it disappears.
• Meeting and call notes: the recap you write immediately after, while it is still fresh.
• Long-form drafting: outlining an essay or argument block by block before you shape it.

Voice Keyboard Pro handles all of these because it does not care that you are in Roam. It types wherever your cursor is, so a graph built out of blocks is just a series of text fields to speak into, one after another.

Dictating in Roam Research on Mac

Roam on the desktop runs in your browser, whether you use the web app or a wrapper. That is the easiest possible case for voice. Voice Keyboard Pro for Mac lives in your menu bar and types at the cursor system-wide, so it works inside a Roam block the same way it works in any native app. There is nothing to install into Roam and nothing to configure inside your graph, and because the app never talks to Roam, it works identically whether your database is hosted, offline, or encrypted.

One-time setup

1. Download Voice Keyboard Pro for Mac from voicekeyboardpro.com and drag it to your Applications folder.
2. Open it once and grant microphone and accessibility permissions when macOS asks. Accessibility is what lets the app place text at your cursor inside the browser.
3. Pick a hotkey you can hold comfortably, such as a function key or a spare modifier.

Dictating into a block

1. Open your graph and click into a block so the cursor is blinking.
2. Hold your hotkey, speak the block, then release.
3. Your words appear in the block within about a second. Press Enter to start the next block, then hold the hotkey and keep going.

Because the app is system-wide, the same hotkey that fills a Roam block also dictates into your email, your terminal, and every other app you use. You build one habit and it follows you everywhere. If you also keep notes in other tools, the same motion works when you dictate in Logseq or across Notion and Obsidian, so switching between graphs never means relearning how to capture.

Dictating in Roam Research on iPhone

Roam does not have a heavy native iOS app the way some note tools do; most mobile capture happens through Safari or a saved web app on the home screen. That is fine for voice, because Voice Keyboard Pro installs as a custom iOS keyboard with a built-in mic button, and a keyboard works in any text field the browser presents. Whether you open Roam as a full page or a mobile shortcut, the mic is right there.

One-time setup

1. Install Voice Keyboard Pro from the App Store.
2. Open Settings → General → Keyboard → Keyboards → Add New Keyboard and add Voice Keyboard Pro.
3. Tap it in the list and enable Allow Full Access. This is what lets the keyboard send audio for transcription and return your text. If you want the reasoning behind that toggle, see our guide on enabling Full Access safely.

Dictating on the go

1. Open Roam in Safari and tap into a block so the keyboard appears.
2. Switch to the Voice Keyboard Pro keyboard using the globe key.
3. Tap the mic button, speak your block, and watch the words appear. Tap again to stop, then move to the next block.

Mobile capture is where a graph either grows or stalls. The whole promise of Roam is that a thought you have while walking to lunch ends up in the same database as everything else, but that only happens if capturing it on your phone is easy. Speaking a block into Roam on your phone takes a few seconds and no thumb gymnastics, which is the difference between capturing the idea and letting it go.

Working with blocks and bullets by voice

Roam's outliner structure is the one thing that makes dictating into it a little different from dictating into a plain document. Every line is its own block, and the structure comes from Enter, Tab, and Shift-Tab rather than from your words. The reliable pattern is to let voice do the prose and let the keyboard do the structure.

• Speak one block, then press Enter. Dictate a complete thought, release the hotkey or tap the mic off, press Enter to open the next block, and speak again. One block per breath keeps your graph clean and your references precise.
• Indent with Tab after you speak. Get the words down first, then use Tab and Shift-Tab to nest blocks under their parent. Trying to voice the indentation is more trouble than it is worth; the keys are instant.
• Dictate long thoughts, then split them. If a spoken block runs long, capture it in full and break it into child blocks afterward. Editing spoken text into structure is faster than composing structured text from scratch.

The habit that works best is simple: talk in blocks. Because your eyes stay on the screen while you speak rather than on your thumbs or your keyboard, it is easy to watch the outline take shape and press Enter at the natural break.

Page links, tags, and block references

The connective tissue of Roam is [[page links]], #tags, and block references, and those are the one place where dictation asks for a light touch. They are more markup than prose, and they depend on Roam's autocomplete popping up as you type the opening brackets. Speaking the brackets character by character is the wrong tool. Speak the words, add the links with a tap.

• Dictate the sentence, then link the terms. Say the block in plain language, then go back and wrap the important terms in double brackets so Roam's autocomplete offers the matching page. Because linking is a deliberate act of curation anyway, doing it as a second pass often produces a better-connected graph than trying to link on the fly.
• Type the brackets, let autocomplete finish. When you do add a page link or tag, type the [[ or # and let Roam's suggestion list do the rest. That guarantees you land on the exact existing page rather than accidentally creating a near-duplicate.
• Keep casing consistent with Smart Vocabulary. If you have project names or people who show up constantly, Smart Vocabulary can make sure they come out spelled and capitalized the same way every time, which matters in a graph where ProjectAtlas and Project Atlas would become two different pages. Our overview of custom vocabulary that learns your words explains how that works.

The general principle holds across every outliner and note app: dictation is best at the words, and the graph structure is best added with a couple of deliberate keystrokes. Let each do what it is good at and the two combine into something faster than either alone.

The daily-notes capture habit

Most serious Roam users live in the Daily Notes page. It is the default landing spot, the running journal, and the inbox for everything that has not found its permanent home yet. It is also the page that benefits most from voice, because its whole job is high-volume, low-friction capture.

The habit to build is talking to your daily note the way you would talk to yourself. You finish a call and speak a three-block recap. You read an article and dictate the one idea worth keeping. You have a thought in the shower and, the moment you are at your phone, you say it into today's page before it is gone. None of these are worth the friction of typing, which is exactly why so many of them never get captured. Spoken, they cost a few seconds each, and a week of them adds up to a graph that actually reflects your thinking rather than the subset of it you had the patience to type.

If you keep a journal habit alongside your research graph, the same approach carries over directly; our piece on voice journaling on Mac covers the daily-capture rhythm in more depth.

Literature notes and drafting by voice

Roam is a favorite of people doing serious reading and writing, and both are natural fits for dictation. A literature note is your own restatement of what a source said, in your own words, which is precisely the prose that voice handles beautifully. Instead of copying quotes, you talk through what the argument was and why it matters, block by block, and you end up with notes you actually understand rather than passages you merely highlighted.

Drafting works the same way. When an essay is still forming, dictating the argument block by block lets you get the whole shape down while it is fresh, then rearrange and link the blocks into a structure afterward. Speaking the first draft and editing it into order is faster and looser than typing a tidy draft from the start, and it tends to produce writing that sounds more like you. If your writing lives partly in longer documents, the same instinct applies when you use voice to improve first drafts anywhere.

Fixing mistakes without breaking flow

No dictation is perfect, and in a graph where a wrong word can quietly become a wrong page link, corrections matter. The good news is that fixing text is quick. On iPhone, Voice Keyboard Pro includes Voice Edit: instead of nudging the cursor between two characters with your thumb, you speak the change you want and the text updates in place. If a term came out wrong, you say what it should be and it is fixed before you ever wrap it in brackets.

On Mac, corrections are usually a matter of clicking into the spot and dictating the replacement, or simply typing it, since your hands are already on the keyboard. Most people settle into a hybrid where voice does the bulk of each block and the keyboard does the small surgery and the structure. That mix is what makes dictation feel like a speed-up rather than a trade-off.

Multilingual notes and research

Plenty of Roam users read and think across languages, especially researchers working with sources that are not in English. On iPhone, Voice Keyboard Pro includes two-way translation while you dictate, across 24 languages. You can speak a note in your own language and have it land in another, or read a source in one language and capture your summary in the language you write in.

That turns a bilingual research workflow from a copy-and-paste chore into a single spoken block. If you regularly work across two languages, it is worth seeing how bilingual voice typing fits into the rest of your day, not just your graph.

A note on privacy

People who build a second brain in Roam tend to care a great deal about where that thinking lives, so it is worth being clear. Voice Keyboard Pro is built so that your dictation content is not stored on our servers. The transcription engine processes your audio to return text, and what we keep on the backend is limited to operational pings, not your audio and not the transcript of what you said. Your graph is yours, and nothing you speak into it is retained by us.

If a private-by-default approach to dictation matters to you across every app, our overview of private voice to text on Mac explains it in more detail.

Free tier and Pro

Voice Keyboard Pro has a free tier with a generous daily allowance, which is plenty to try dictating your daily notes and a few literature notes and decide whether it fits how you use Roam. If you capture into your graph throughout the day and want voice everywhere, Pro removes the daily limits and unlocks the full feature set, including two-way translation, for $4.99 per month or $34.99 per year. Pro covers both the Mac app and the iPhone keyboard, so one subscription follows you from your desk to your phone.

Frequently asked questions

Does dictating in Roam Research need a plugin or extension?

No. On Mac, Voice Keyboard Pro types at the cursor across the whole system, so it works inside a Roam block in any browser with nothing to install into your graph. On iPhone, it is a keyboard, so it works wherever you open Roam, including Safari and a home-screen web app.

Will it work with my hosted or offline graph?

Yes. Because the app never connects to Roam and only places text at your cursor, it does not matter whether your database is hosted, offline, or encrypted. To Voice Keyboard Pro, a block is just a text field.

How do I dictate page links and tags?

Dictate the words in plain language, then add [[page links]] and #tags with a tap so Roam's autocomplete matches them to existing pages. Adding links as a second pass keeps your graph free of accidental duplicate pages.

Can I capture notes in another language?

Yes. On iPhone, Voice Keyboard Pro offers two-way translation across 24 languages while you dictate, so you can read a source in one language and capture your summary in another.

Is what I dictate private?

Your dictation content is not stored on our servers. The backend keeps operational pings only, not your audio or your transcripts, so everything you speak into your graph stays yours.

Start dictating into your graph

The next time a thought is worth keeping, try speaking it into Roam instead of typing it. Click into a block or tap the mic, say what you mean, and let the graph do what it is good at once the words are down. The observation you would have skipped gets captured, your daily notes fill up with the things you actually thought, and your second brain starts to reflect the full width of your thinking. Download Voice Keyboard Pro and dictate your next block to feel the difference.