You are reading, not composing
Both rounds give you the words, just like the study. Real writing adds thinking time to both methods, so the gap usually narrows when you have to decide what to say.
This speaking vs typing test takes two rounds: type a short passage, then read the same passage aloud. Both rounds are scored in words per minute on the same text, and your result is set against the research behind the famous “3x faster” claim.
Short answer: speaking is roughly three times faster than typing on a phone. In the peer-reviewed study behind the claim, 24 English speakers dictated at 152.9 words per minute and typed at 52.2. For comparison, adults average about 183 words per minute speaking when they read aloud, 196 in phone conversation, and type at 51.6 on a physical keyboard.
The test measures your typing speed and your speaking speed on the same passage of 55 to 59 words. Pick a passage, type it at your normal pace, then read it aloud the way you would dictate a message. Changing the passage restarts both rounds so the comparison stays fair.
The clock starts on your first key and stops on your last. Type at your normal pace; you can fix mistakes as you go.
Thanks for sending the draft over this morning. I read it on the train and I think the structure works well. The opening could be shorter, and the second section needs one clear example. If you can make those changes by Friday, I will share it with the rest of the team and we can plan the launch together.
Almost every page that says speaking is three times faster than typing traces back to a single Stanford-led experiment. The headline number is real, but it changed between versions, and the conditions matter.
32 people (16 English, 16 Mandarin speakers). English: 161 words per minute by voice, 53 on the keyboard.
arXiv 1608.07323, version 1The Stanford News story that most pages cite. It reports the preprint’s 32-person numbers and quotes Sherry Ruan, the paper’s first author and then a Stanford PhD student: “we were actually quite surprised to find that it was almost three times faster than typing on a keyboard.”
Stanford NewsRe-run with 48 people, 24 per language. English: 152.9 vs 52.2 words per minute. Mandarin: 123.0 vs 42.8 (2.87x).
doi.org/10.1145/3161187The number travels on its own, usually without the sample size, the phone, the task or the newer 2.93.
| Version | People | Voice, wpm | Keyboard, wpm | Voice vs keyboard |
|---|---|---|---|---|
| 2016 preprint (English) | 16 of 32 | 161.2 | 53.5 | 3.01x |
| 2017 peer-reviewed (English) | 24 of 48 | 152.9 | 52.2 | 2.93x |
Adults type about 52 words per minute on a physical keyboard and 36 on a phone, read aloud at about 183 and talk at about 196. Each number below is checked against the paper it comes from, with who was measured.
| What | Words per minute | Who was measured | Source |
|---|---|---|---|
| Typing on a phone | 36.2two thumbs: 38.0 | 37,370 volunteers, online typing test | Palin et al., MobileHCI 2019 |
| Typing on a physical keyboard | 51.6fastest tenth: above about 78 | 168,960 volunteers, online typing test | Dhakal et al., CHI 2018 |
| Typing on a phone, in the lab | 52.2 | 24 native English speakers, iPhone 6 Plus | Ruan et al., IMWUT 2017 |
| Dictating on a phone, after fixing errors | 152.9 | The same 24 people | Ruan et al., IMWUT 2017 |
| Dictating on a phone, before fixing errors | 178.9 | The same 24 people | Ruan et al., IMWUT 2017 |
| Reading aloud | 183 | 77 studies, 5,965 adults | Brysbaert, Journal of Memory and Language 2019 |
| Talking on the phone | 196236 not counting pauses | 2,438 recorded phone conversations (Switchboard) | Yuan, Liberman and Cieri, Interspeech 2006 |
| Reading silently, non-fiction | 238fiction: 260 | 190 studies, 18,573 adults | Brysbaert, Journal of Memory and Language 2019 |
About “100 to 150 words a minute”. Many guides give 100 to 150 words a minute for presentations and 150 to 160 for audiobooks. We could not trace either range to a primary study, so neither is in the table. The measured averages for reading aloud (183) and phone conversation (196) are both higher.
Use the table freely with a link back: download it as CSV (CC BY 4.0, checked 7 October 2026).
Your ratio depends on four things the 3x study held fixed: what you are writing, how many mistakes you fix, which keyboard you use and whether you speak the punctuation.
Both rounds give you the words, just like the study. Real writing adds thinking time to both methods, so the gap usually narrows when you have to decide what to say.
In the study, dictation ran at 178.9 words a minute before corrections and 152.9 after. A recogniser that mishears names or jargon raises that cost, which is why this test shows accuracy next to speed.
The 3x compared voice with a phone keyboard. On a physical keyboard the average is 51.6 words a minute and the fastest tenth of typists pass about 78, so a fast typist may see closer to 2x.
The study phrases had no punctuation. Saying "comma" and "new paragraph" out loud slows dictation down, unless your dictation tool adds the punctuation and layout for you.
Both rounds use one rule: passage words reached divided by minutes taken, with accuracy reported separately.
Contextli never receives, stores or uploads your audio, and what you type never leaves this page. The speaking round relies on the speech service built into your browser. In Chrome, for example, the audio is sent to Google’s servers for recognition (MDN explains how), unless on-device recognition is available, in which case the test uses it and says so. The stopwatch uses no microphone.
Contextli (7 October 2026). Speaking vs typing speed test, with the published rates behind the "3x faster" claim. https://contextli.com/tools/speaking-vs-typing-test
Short answers, using the same primary studies as the table above.
For most people, yes. In the peer-reviewed study behind the famous claim, 24 native English speakers dictated short phrases on an iPhone at 152.9 words per minute after fixing errors, against 52.2 on the phone keyboard: 2.93 times faster. Against a physical keyboard the gap is smaller, because the average there is 51.6 words per minute and the fastest tenth of typists pass about 78. The test on this page measures your own ratio on the same passage.
It depends on the task. Adults read aloud at about 183 words per minute on average (Brysbaert, 2019, 77 studies). In recorded American phone conversations the rate is about 196 words per minute, or 236 if you leave out pauses (Yuan, Liberman and Cieri, 2006). Dictation in the Stanford-led phone study ran at 152.9 words per minute once errors were fixed.
Yes, 150 is a comfortable, clear pace for dictating or presenting. It is a little slower than the measured averages for reading aloud (183) and phone conversation (196), which is usually a good thing when someone, or a speech recogniser, has to follow every word.
It is fast for reading aloud or presenting, and close to average for casual conversation: recorded American phone calls averaged 196 words per minute, and individual conversations ranged from 111 to 291 (Yuan, Liberman and Cieri, 2006).
On a physical keyboard the average across 168,960 people was 51.6 words per minute; the slowest tenth typed below about 26 and the fastest tenth above about 78. On a phone the average was 36.2 words per minute (37,370 people). So 36 is average on a phone and below average on a computer keyboard.
At the average reading-aloud rate of 183 words per minute, about 1,800 words. If you speak more slowly for an audience, at 130 to 150 words a minute, it is 1,300 to 1,500 words. Time yourself with the stopwatch mode in the test above to get your own number.
From a Stanford-led study by Ruan, Wobbrock, Liou, Ng and Landay. The first version, posted in August 2016 with 32 participants, found English dictation 3.0 times faster than an iPhone keyboard (161 vs 53 words per minute). The peer-reviewed version, published in Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies with 48 participants, found 2.93 times (152.9 vs 52.2). Both measured people copying short given phrases on a phone, not composing their own text.
No. Contextli never receives, stores or uploads your audio, and the text you type never leaves the page. The speaking round uses the speech recognition built into your browser; in Chrome, for example, that audio goes to Google’s servers for recognition unless on-device recognition is available. If you would rather no audio leave your device at all, use the stopwatch option, which does not use the microphone.
The speaking round needs the Web Speech API. Chrome and Safari include it (Safari needs Siri turned on). Firefox does not ship it by default, and some other browsers expose it without it working, so the test offers a stopwatch instead: press Start, read the passage aloud, press Stop. If the microphone is blocked, allow it from the address bar and try again.
Every source was opened and checked on 7 October 2026. Where a paper is behind a paywall, the open copy is linked next to it.
Dictating on a computer? See voice to text on Mac and voice to text on Windows.
Contextli turns what you say into a finished email, message or note, with the punctuation and tone that fit the app you are typing in.