Dictation vs Typing: When Speech-to-Text Wins (and When It Doesn't)

You speak at 130–160 words per minute. The average person types at 40–50. By that arithmetic, dictation should have killed the keyboard a decade ago — and modern speech recognition is genuinely excellent, with word error rates on clean audio down around 5–8%. Yet almost nobody dictates their workday. The arithmetic is missing a term, and that term is editing.

The Headline Numbers

Conversational speech runs about 150 WPM. Deliberate dictation — speaking punctuation, pausing to think in silence rather than with filler words — settles closer to 100–120 WPM of usable draft. Typing spans a much wider range: the population average sits in the mid-40s, a trained professional types 70–90, and you can check where any given speed falls in our breakdown of what counts as a good typing speed. So dictation starts every race with a 30–70 WPM head start. The race just doesn't end at the draft.

The Editing Tax, Itemised

A 5–8% word error rate sounds small until you multiply it out. Dictate a 1,000-word document and you've planted 50–80 errors — and they're not typing-style errors. A mistyped word is usually visible garbage ("teh"); a misrecognised word is a real word that's wrong ("their best ascent" for "their best assent"), which spellcheck won't flag and skim-reading won't catch. Each correction means locating the error, repositioning the cursor (by voice, painfully, or by hand — at which point you've returned to the keyboard anyway), and fixing it: realistically 8–15 seconds per error.

Method	Draft speed	Errors per 1,000 words	Correction time	Effective speed
Dictation (clean audio)	~110 WPM ≈ 9 min	50–80	8–15 min	40–60 WPM
Dictation (open office, jargon)	~110 WPM ≈ 9 min	100–150	15–25 min	30–40 WPM
Typing at 50 WPM, 95% accuracy	20 min	caught mostly in-line	2–4 min	~45 WPM
Typing at 80 WPM, 97% accuracy	12.5 min	caught mostly in-line	1–3 min	~70 WPM

The pattern is stark: dictation's effective speed for clean, finished text collapses to roughly 40–60 WPM. That beats an average typist. It loses, clearly, to a good one.

Where Dictation Genuinely Wins

• Long first drafts where polish comes later. If you're going to rewrite heavily anyway, the editing tax was already budgeted. Authors dictating 5,000-word rough chapters are playing dictation's best game.
• Hands-busy and away-from-desk capture. Voice notes while driving or walking have no typing competitor at all — the alternative isn't typing, it's losing the thought.
• RSI, injury, and accessibility. For typists managing repetitive strain, dictation isn't a productivity hack, it's continuity of career. This is the use case where the comparison stops mattering.
• Short, low-stakes messages. A two-sentence chat reply has maybe one error to fix. The editing tax rounds to zero.

Where Typing Stays Faster — Probably Forever

Editing is the big one. Revision is surgical: jump to a word, swap it, delete a clause, restructure. Voice commands for cursor control ("select the previous sentence... no, the one before") are slower than arrow keys by an order of magnitude, which is why every serious dictation workflow still ends at a keyboard.

Code is unwinnable for speech. Try saying useState, user_id, or a regex aloud. Identifiers, brackets, and operators have no natural spoken form, and recognition models are trained on prose. The symbol-dense typing that a coding typing test measures is exactly the typing speech-to-text handles worst.

Precision text: legal language, numbers, names, anything where "close" is wrong. And shared spaces: open offices and homes make dictation socially expensive, and confidential content makes it risky out loud.

Finding Your Personal Break-Even

The decision reduces to one comparison: your sustained typing speed versus dictation's effective 40–60 WPM. Note the word sustained — a 30-second burst score flatters you. Take a 5-minute typing test, which is long enough for fatigue and real-word variety to show up. If you sustain under 50 WPM, dictation will beat you for drafts today, and learning to dictate well is rational. If you sustain 70+, typing wins for nearly everything except long rough drafts and away-from-desk capture.

The strongest workflow most people land on is the hybrid: dictate the messy first pass when the material suits it, then edit and finish by keyboard. Both skills compound — and unlike the speech models, your typing speed is the one variable in this equation you can directly train.