What is a speech jammer and how does it work?

It plays what you say back to you a fraction of a second late. Hearing your own voice delayed, known as delayed auditory feedback, makes many people slow down, stretch sounds, repeat syllables or stop mid-sentence, because people steer their speech partly by hearing it. The effect has been studied since the 1950s, and a 2012 device built on it, the SpeechJammer, won an Ig Nobel Prize. This page does the same job with your microphone and headphones: the default delay is 200 ms and you can set any whole number of milliseconds from 50 to 500 ms, with the slider or by typing, while you talk.

Why do I need headphones?

Because a microphone hears speakers. Through speakers, the delayed sound goes back into the microphone, is delayed and played again, and can build into a loud echo or howl. Headphones keep the delayed voice in your ears only. The page asks the browser to switch echo cancellation off so your voice comes back untouched; with it off, nothing holds back the sound from speakers. To keep a mistake gentle, the volume opens at 25% on every visit and fades in from silence over half a second each time you press Start.

The volume runs from 0 to 100%, and it is capped at 100%: at that setting your voice comes back at the level the microphone delivered it, and the page never amplifies it beyond that. Typing a bigger number holds it at 100%, and the page says so. The page also asks the browser to switch automatic gain control off, so the browser is asked not to boost a quiet microphone either. How loud 100% sounds still depends on your device's own volume, which the page cannot see or change, so it is worth turning that down before you start.

How accurate is the delay?

The delay is applied by the browser's own delay line to the nearest audio sample: 200 ms at 48 kHz is 9,600 samples, and 50 ms at 44.1 kHz is 2,205. The readout shows that count at your device's sample rate once a session starts. What you hear is that delay plus your device's own latency, the time the microphone signal takes to reach the browser and the sound takes to reach your ears. The Heard after line shows the part of that latency the browser reports, which can leave some out, such as the extra delay of wireless headphones. Wired headphones usually add less.

Is my voice recorded or uploaded?

No. The sound goes from your microphone through the browser's delay line to your headphones and nowhere else: nothing is recorded, saved or sent. Press Stop, press Escape, switch to another tab or leave the page, and the page stops the microphone, disconnects the audio and closes the audio engine, so the browser's recording indicator for this page goes off. If you press Stop while the permission prompt is still open, a microphone you allow afterwards is closed again at once. The only thing remembered on this device is the delay you last chose; the volume is never remembered, and Start over at the top of the page forgets the delay too.

What if it says the microphone is blocked or missing?

The browser asks for the microphone only after you press Start, and the page then says what went wrong. If access is blocked, allow the microphone from the icon in the address bar and press Start again. If no microphone is found, or another app such as a call is holding it, the message names that instead. The page needs a secure (https) connection and a current version of Chrome, Edge, Firefox or Safari, and if the browser will not start audio playback the session is ended and the microphone released rather than left open with nothing playing.

Is this a speech therapy tool?

No. It is an entertainment effect and makes no therapeutic claim. Delayed auditory feedback also appears in speech research and clinical settings, but this page is not designed, tested or offered for any speech condition, and its delay, level and timing have not been checked against any clinical device.

Can I use it without a mouse?

Yes. Start and Stop are one large button that stays in the same place, so keyboard focus stays on it when it changes, and Escape stops a running session from anywhere on the page. Each slider moves with the arrow keys and has a number field; a value the page cannot use is explained, and the last valid value stays in use. On a phone the Stop button is pinned to the bottom of the screen while a session runs, and the sliders, fields and buttons are all at least 44 pixels tall.

Speech Jammer

Speech Jammer plays your own voice back into your headphones a moment late, which makes fluent talking surprisingly hard for many people. Press Start, allow the microphone and speak: the delay is 200 ms by default and can be set anywhere from 50 to 500 ms while you talk, and the page plays only the delayed copy, with no direct sound mixed in. The volume starts at 25% on every visit, fades in from silence over half a second each time you press Start, and is capped at 100%, the level your microphone captured, so the page never amplifies your voice. A large Stop button shows how long the session has run, and Stop, Escape, switching tabs or leaving the page all close the microphone and the audio engine. Nothing is recorded or uploaded: the sound goes from the microphone through the browser's delay line to your headphones and nowhere else.

Your input

Put on headphones first. Through speakers the microphone hears its own delayed output, which can build into a loud echo or howl. Volume starts at 25%, fades in from silence over half a second, and is capped at 100%: the level your microphone captured, never louder.

Delay (50 to 500 ms)

Volume (0 to 100%)

Your browser asks for the microphone only after you press Start.

Result

Chain: microphone (off), delay line 200 ms, gain 25%, capped at 100%, then your headphones. An impulse you make at 0 ms is heard 200 ms later at 25% of its level, plus your device's own latency. There is no direct path: you hear only the delayed voice.

00:00Sessionmicrophone off

Delay: 200 ms = 9,600 samples at 48 kHz
Volume: 25% (cap 100%)
Session: 00:00
Microphone: Off: asks when you press Start
Heard after: 200 ms set, plus device latency (shown after Start)

The microphone is off. Nothing is heard until you press Start and allow it.

Delay lineimpulse at sample 0, 200 ms at 48 kHz
impulse moves to sample 9,600
Shortest delay50 ms at 44.1 kHz
2,205 samples
Gain capvolume requested at 150%
held at 100%: gain 1, the captured level, never louder

Press Start, allow the microphone and talk: the page plays only the delayed copy of your voice, 200 ms late with the current setting, plus the latency of your own device. Both sliders work while you speak, move with the arrow keys and have a number field that takes typed values. Every Start fades in from silence over 500 ms. The session timer runs while the microphone is live; Stop, Escape, switching tabs or leaving the page all stop the microphone and close the audio engine. Nothing is recorded or uploaded, and only the delay is remembered on this device.

Common questions

What is a speech jammer and how does it work?: It plays what you say back to you a fraction of a second late. Hearing your own voice delayed, known as delayed auditory feedback, makes many people slow down, stretch sounds, repeat syllables or stop mid-sentence, because people steer their speech partly by hearing it. The effect has been studied since the 1950s, and a 2012 device built on it, the SpeechJammer, won an Ig Nobel Prize. This page does the same job with your microphone and headphones: the default delay is 200 ms and you can set any whole number of milliseconds from 50 to 500 ms, with the slider or by typing, while you talk.
Why do I need headphones?: Because a microphone hears speakers. Through speakers, the delayed sound goes back into the microphone, is delayed and played again, and can build into a loud echo or howl. Headphones keep the delayed voice in your ears only. The page asks the browser to switch echo cancellation off so your voice comes back untouched; with it off, nothing holds back the sound from speakers. To keep a mistake gentle, the volume opens at 25% on every visit and fades in from silence over half a second each time you press Start.
How loud can it get?: The volume runs from 0 to 100%, and it is capped at 100%: at that setting your voice comes back at the level the microphone delivered it, and the page never amplifies it beyond that. Typing a bigger number holds it at 100%, and the page says so. The page also asks the browser to switch automatic gain control off, so the browser is asked not to boost a quiet microphone either. How loud 100% sounds still depends on your device's own volume, which the page cannot see or change, so it is worth turning that down before you start.
How accurate is the delay?: The delay is applied by the browser's own delay line to the nearest audio sample: 200 ms at 48 kHz is 9,600 samples, and 50 ms at 44.1 kHz is 2,205. The readout shows that count at your device's sample rate once a session starts. What you hear is that delay plus your device's own latency, the time the microphone signal takes to reach the browser and the sound takes to reach your ears. The Heard after line shows the part of that latency the browser reports, which can leave some out, such as the extra delay of wireless headphones. Wired headphones usually add less.
Is my voice recorded or uploaded?: No. The sound goes from your microphone through the browser's delay line to your headphones and nowhere else: nothing is recorded, saved or sent. Press Stop, press Escape, switch to another tab or leave the page, and the page stops the microphone, disconnects the audio and closes the audio engine, so the browser's recording indicator for this page goes off. If you press Stop while the permission prompt is still open, a microphone you allow afterwards is closed again at once. The only thing remembered on this device is the delay you last chose; the volume is never remembered, and Start over at the top of the page forgets the delay too.
What if it says the microphone is blocked or missing?: The browser asks for the microphone only after you press Start, and the page then says what went wrong. If access is blocked, allow the microphone from the icon in the address bar and press Start again. If no microphone is found, or another app such as a call is holding it, the message names that instead. The page needs a secure (https) connection and a current version of Chrome, Edge, Firefox or Safari, and if the browser will not start audio playback the session is ended and the microphone released rather than left open with nothing playing.
Is this a speech therapy tool?: No. It is an entertainment effect and makes no therapeutic claim. Delayed auditory feedback also appears in speech research and clinical settings, but this page is not designed, tested or offered for any speech condition, and its delay, level and timing have not been checked against any clinical device.
Can I use it without a mouse?: Yes. Start and Stop are one large button that stays in the same place, so keyboard focus stays on it when it changes, and Escape stops a running session from anywhere on the page. Each slider moves with the arrow keys and has a number field; a value the page cannot use is explained, and the last valid value stays in use. On a phone the Stop button is pinned to the bottom of the screen while a session runs, and the sliders, fields and buttons are all at least 44 pixels tall.

The delay you set is applied to the nearest audio sample; your microphone, audio system and headphones add their own latency on top, and the page can show only the part the browser reports. It is an entertainment effect and makes no therapeutic claim. Through speakers the delayed sound feeds back into the microphone, so it is made for headphones.