KeyslamKeyboard sound test
Keyboard sound test: compare 15 real switches and typewriters
Play two boards back to back with the same phrase, the same rhythm and matched loudness, then try a blind round. Below that: measured loudness and brightness for every board, quietest to loudest.
Updated
Short answer: at medium strength the 15 boards measure within 2.2 LU of each other (-28.7 to -26.5 LUFS), because every pack is leveled to the same loudness. The differences show when you hit hard: Topre and Unicomp stay the quietest at slam (-23.1 and -22.9 LUFS), and Vintage Technics and Olivetti get the loudest (-20.5 and -20.0 LUFS). Use headphones and the A/B player below to hear the rest: the click, the thock and how long each key rings.
A/B player
Keyboard sound tester: two boards, one phrase
Both boards play "the quick brown fox jumps over the lazy dog" with identical timing, at the strength you pick. Each key always plays its own take, so a repeat plays the same strike.
Pick two boards, then press Play. The first play of a board takes a moment to load.
Keyboard sound test: compare 15 real boards by loudness
Sorted from the quietest to the loudest board when you strike as hard as you can. At medium, the strength most typing lands on, they measure within 2.2 LU of each other.
| Board (press to play) | Type | Medium (LUFS) | Medium peak (dBTP) | Slam (LUFS) | Soft to slam (LU) |
|---|---|---|---|---|---|
| Electro-capacitive | -26.5 | -6.1 | -23.1 | 9.9 | |
| Buckling spring | -27.7 | -7.7 | -22.9 | 9.3 | |
| Buckling spring | -28.1 | -8.9 | -22.9 | 9.8 | |
| Tactile | -27.6 | -4.8 | -22.5 | 9.9 | |
| Linear | -27.7 | -6.1 | -21.8 | 10.7 | |
| Clicky | -27.4 | -6.1 | -21.6 | 11.3 | |
| Tactile | -27.3 | -8.6 | -21.6 | 10.5 | |
| Mechanical | -27.7 | -4.1 | -21.6 | 11.2 | |
| Typewriter | -28.7 | -10.0 | -21.5 | 12.5 | |
| Linear | -27.5 | -6.2 | -21.4 | 9.4 | |
| Tactile | -27.4 | -6.5 | -21.3 | 10.2 | |
| Linear | -27.4 | -7.9 | -21.2 | 11.7 | |
| Butterfly | -27.4 | -6.1 | -20.8 | 11.5 | |
| Linear, modded | -27.6 | -5.0 | -20.5 | 11.3 | |
| Typewriter | -27.8 | -8.0 | -20.0 | 13.3 |
Measured by us on November 2, 2026 from the sound files this page plays, at 100% volume. LUFS is perceived loudness (closer to zero is louder); dBTP is the highest true peak; LU is the gap in loudness between the soft and slam bands. Method below.
What makes each board sound different
Laptop, Rainy 75 and Topre keep the most energy below 200 Hz, the low thump people call thock. Olivetti, Unicomp and Model F keep the most above 2 kHz, where clicks and typewriter strikes live.
| Board | Below 200 Hz | Above 2 kHz | Brightness (centroid) | In a sentence |
|---|---|---|---|---|
| 90% | 2% | 253 Hz | A 2018 MacBook Pro's butterfly keys, close-miked: soft and subtle. | |
| 90% | 4% | 264 Hz | A Rainy 75 keyboard (switches unstated): a dark, creamy thock. | |
| 33% | 6% | 803 Hz | Cherry MX Brown switches: a soft bump and a muted clack. | |
| 67% | 11% | 819 Hz | A Happy Hacking Keyboard with Topre switches: deep, rounded thock. | |
| 21% | 9% | 1,152 Hz | An old Technics board with waxed Gateron Yellows: chunky and old-school. | |
| 6% | 20% | 1,238 Hz | Gateron Red switches on a GMMK TKL: dark, smooth and a little hollow. | |
| 33% | 16% | 1,338 Hz | Cherry MX Black switches on a Leopold FC980M PD: heavy, smooth and full-bodied. | |
| 55% | 29% | 1,474 Hz | Leopold FC660M, MX Browns and PBT caps, close-miked: crisp and detailed. | |
| 26% | 25% | 1,492 Hz | A WhiteFox board with Hako Violet switches: punchy, with a long ring. | |
| 13% | 28% | 1,650 Hz | Cherry MX Red switches on a Corsair K95: smooth, light and quick. | |
| 20% | 46% | 1,876 Hz | Cherry MX Blue switches: a crisp click on every press. | |
| 22% | 50% | 2,704 Hz | An early-1980s IBM Model F XT: buckling springs with a loud, deep clack. | |
| 5% | 32% | 3,447 Hz | IBM Selectric II: the electric typeball strike, with a bell on Return. | |
| 8% | 61% | 3,507 Hz | A Unicomp buckling-spring keyboard, the Model M design: a big, springy click-clack. | |
| 5% | 82% | 5,017 Hz | Olivetti Lettera 22: a manual portable's snappy typebars and a bell. |
Sorted from deepest to brightest. Shares are of each strike's energy, and brightness is its spectral centroid: the higher the number, the brighter the strike. Each board's own page lists its decay and peak range. For the vocabulary, see thocky vs clacky, explained.
Membrane / laptop vs mechanical keyboard sound test
Most people asking this want to know what they'd gain by moving from a flat laptop or membrane keyboard to a mechanical one. The closest thing to a membrane board here is the Laptop pack: a 2018 MacBook Pro butterfly keyboard, close-miked, with short travel, thin caps and no mechanical switch. The mechanical side is a Cherry MX Brown, a tactile switch with no click, which keeps the comparison about the switch and not about a loud click jacket.
The numbers agree with your ears. The laptop strike keeps 90% of its energy below 200 Hz and 2% above 2 kHz; the MX Brown keeps 33% and 6%. At slam the laptop measures -20.8 LUFS and the MX Brown -21.6 LUFS, so what you hear between them is mostly character, not volume.
Listen for three things: the laptop has a short, dull tap with almost no top end; the MX Brown adds a plastic clack as the cap bottoms out; and the MX Brown rings a little longer. To hear the laptop recording key by key, open the MacBook keyboard sound test.
How to run a fair switch sound test
- Use headphones and set the volume once. Changing it between boards makes the louder one sound "better".
- Keep the strength the same. The A/B player locks it. The Play buttons in the tables vary each press a little, like normal typing, so use the A/B player for a strict comparison.
- Listen for one thing at a time: the click before bottom-out (only clicky switches have it), the pitch of the bottom-out, the ring after the key, and the sound on the way back up.
- Go blind. Pick two close boards, like two tactile ones, and play X ten times. If you score around 5 of 10, you probably can't hear a difference, and that's useful to know before you buy a keyboard.
How we measured
On November 2, 2026 we rendered the same 40-key passage from each board's soft, medium, hard and slam strikes: the letters of “the quick brown fox jumps over the lazy dog”, repeated to 40 keys, one every 150 ms (about 80 words per minute) with its key-up 80 ms later, at 100% volume, the way this page plays them, minus its small random variations. Each passage was read with ffmpeg's EBU R128 meter for integrated loudness and true peak. The frequency numbers come from the first 85 ms of every medium and hard strike, median across strikes. All 15 boards were measured the same way on the same day.
The packs in the Keyslam keyboard sound app for Mac are cut from one real typing recording each (the typewriters add a bell recording for Return). Keyslam sorts the strikes by natural loudness into soft, medium, hard and slam (the hardest ~8%) and levels each band, and every pack's standard typing passage is leveled to the same loudness (-28 LUFS). That is why the medium column is so even. More on the recordings is on the Keyslam about page.
Keyboard sound test questions
Which keyboard sound is the quietest?
At medium strength the 15 boards measure within 2.2 LU of each other (-28.7 to -26.5 LUFS), because every pack is leveled to the same loudness. The differences show when you hit hard: Topre and Unicomp stay the quietest at slam (-23.1 and -22.9 LUFS), and Vintage Technics and Olivetti get the loudest (-20.5 and -20.0 LUFS).
Why are the boards matched at medium but not at slam?
Because the leveling targets medium. Keyslam sets every pack so a standard medium-strength typing passage measures -28 LUFS, which is why the medium column sits within 2.2 LU. Within each band, every take is moved most of the way, not all the way, toward the band's level target, so part of each recording's natural spread survives, and that spread shows in the slam column. At medium, this is a test of character: click, thock, ring and the key-up sound.
What do LUFS, LU and dBTP mean?
LUFS (loudness units relative to full scale) measures perceived loudness, with numbers closer to zero being louder. A difference of 1 LU equals 1 dB. dBTP is the true peak, the highest point of the waveform, where 0 is the top before clipping.
Is a laptop keyboard sound test useful if I already own a MacBook?
Yes. The Laptop pack is a recording of a real 2018 MacBook Pro butterfly keyboard, so it shows what a gentle "just confirm the press" sound is like next to a mechanical board, at the same loudness.
Found your board? Hear it on every keystroke.
Keyslam plays all 15 boards in all your Mac apps for $3.99 once. Try it free for 7 days.