Reverse audio: play any sound backwards, right in your browser

Reversed speech doesn't sound like words. It has the same voice and the same rhythm of syllables, but sounds that normally start sharply, such as a “t” or a “k”, now fade in and stop suddenly. Vowels hold up better, which is why it can feel almost like a foreign language.

That's also why a reversed sentence isn't simply the words in the opposite order. Each sound inside each word is backwards too, so “hello” doesn't become “olleh” as you'd read it.

The most common uses:

• Sound design — a reversed hit makes a rising “whoosh” into a scene change or a drop.
• Reverse cymbals — a crash played backwards swells up to a peak, so it can lead into the next section of a song.
• Reverse reverb — reverse a sound, add reverb in an audio editor, then reverse it again, so the echo arrives before the sound.
• The reverse singing challenge — copy a backwards phrase, record it and reverse it to see how close you got.
• Curiosity — checking what a song or a word really sounds like the other way round.

Backmasking means recording a sound backwards onto a track that's meant to be played forwards. It's real as a studio technique: The Beatles popularised it, and Electric Light Orchestra's “Fire On High” contains a deliberate backwards message ^[1].

The claim that songs carry secret backwards messages that influence listeners is a different thing. As Wikipedia summarises the research, Vokey and Read found no effect from the meaning of backward messages on listeners' behaviour ^[1], and in a study of 40 people, listeners couldn't hear the alleged phrases unless they were prompted first ^[2]. Hearing words in reversed sound is usually pareidolia: the brain finding patterns in noise. This tool shows you what's on the recording, and it can't tell you what anyone meant by it.

When you choose a file, the browser reads it from your disk and decodes it into raw audio in memory, using the Web Audio API ^[3]. Recordings are captured the same way, with the browser's MediaRecorder ^[4]. The tool then reverses the samples of each channel. No part of the audio is sent anywhere, and closing the tab clears it.

Voice clean-up filters are switched off while recording, because they can cut out held notes and quiet sounds. That means background noise stays in, so a quiet room helps.

We timed the tool on 4 November 2026 in Edge on one Windows desktop, using generated test tones. Your times will differ, and we haven't measured phones or other browsers.

• One-hour MP3, 57.6 MB at 128 kbps — opened and reversed in 8.9 seconds. Exporting took 1.8 seconds and made a 691.2 MB WAV.
• 5 minute 40 second WAV, 60 MB — opened in 0.8 seconds and exported in 0.2 seconds.
• A full 60-second recording — stopped on its own at the limit and came to an 11.5 MB WAV.
• Anything over 60 MB — refused with a message, before it's read.

Browsers can decode MP3 but don't come with a way to write one, so an MP3 export would need an extra encoder inside the page. This tool doesn't have one. WAV is plain uncompressed audio, which also means the reversed sound isn't compressed a second time.

The catch is size: about 10.6 MB per minute for stereo at 44.1 kHz, and about 11.5 MB at 48 kHz. The tool saves at your browser's audio rate and shows the size before you download.