Reverse Audio performs the entire reverse operation locally in your browser tab, which means the MP3, WAV, M4A, AAC, Ogg, or WebM file you choose is not uploaded to a remote server before, during, or after processing. The browser validates the input file size (up to 50 MB), reads the bytes into an ArrayBuffer, decodes them with the Web Audio API, reverses every sample in every channel, encodes the result as a PCM16 WAV, and exposes it through a temporary object URL for preview and download. No codec conversion server is called, no FFmpeg binary is installed on your device, and no job queue sends the data off your machine. The original file bytes, decoded samples, filename, and finished WAV stay within the current page. The temporary URLs that drive the built-in preview and the download are released automatically when you pick another file, run the operation again, or close the page, and the Web Audio context used for decoding is closed after the operation as well. Because nothing travels to a remote endpoint, the result is reachable even on connections with restricted outbound data transfer once the page itself has loaded.
What "Local Browser Processing" Actually Means
A reverse audio tool can work in two fundamentally different ways. The first reads your file on a remote server, runs a command-line utility like FFmpeg or SoX there, and streams the result back to your browser. The second loads the file into the current browser tab, decodes it through the Web Audio API specification for AudioBuffer, manipulates the resulting buffer in JavaScript, and encodes the output through JavaScript as well. The Reverse Audio tool follows the second path, and that path is the reason the question of uploading does not arise in the first place. According to the underlying Web Audio behavior, decoding is performed by the browser's audio engine on the bytes you provide, producing a buffer whose sample rate and channel count match the source. Reversing channels over that buffer is purely a local array operation and never needs a network round trip.
This local workflow has concrete consequences for the user. There is no upload progress bar because nothing is in transit. There is no server-side job queue where a file could be cached after you close the tab. The page never asks for a login because no account is needed to receive your data. You can preview, reverse, and download on a connection that blocks arbitrary outbound traffic once the page assets themselves are cached. Processing speed depends on the CPU in your own machine rather than on shared server load, which is why long decodes can feel slower on older phones than on a modern laptop.
File Types and Explicit Limits
Acceptance happens in two stages inside the file picker. First, the picker checks the MIME type to confirm the container is one of the formats the tool targets — MP3, WAV, M4A, AAC, Ogg, or WebM. Second, the Web Audio API tries to decode the bytes, and a particular browser or operating system may not include every codec. A familiar container can also hide an unsupported codec, so an Ogg file in one browser may decode perfectly while the same file in another browser raises a decode error. When decoding fails, the tool surfaces that error rather than silently producing a result from different data.
Five explicit numeric limits guard the rest of the operation. The compressed input must be 50 MB or smaller. The decoded audio must be five minutes or shorter. The decoded buffer must contain between one and eight channels. The sample rate must fall between 8,000 and 192,000 Hz. The total channel-sample budget — frames multiplied by channels — must stay under 30 million. That last budget protects the tab from unexpectedly large arrays even when a compressed source looks small. A file crossing any boundary receives a specific message and is not partially processed, so a successful download cannot come from input that should have been rejected. Changing files or retrying clears the previous download first, so an old success cannot be confused with a fresh result after an error.
| Limit | Value |
|---|---|
| Input compressed size | up to 50 MB |
| Decoded duration | up to 5 minutes |
| Channel count | 1 to 8 |
| Sample rate | 8,000 to 192,000 Hz |
| Channel samples total | up to 30,000,000 |
Reverse an Audio File Locally
Running the Reverse Audio tool takes a few clear steps and leaves the audio on your machine the whole time.
- Open the Reverse Audio tool in your browser tab.
- Click the file picker and choose an MP3, WAV, M4A, AAC, Ogg, or WebM file from your device. Confirm the file is 50 MB or smaller before selecting it, since the picker enforces that cap on the raw bytes.
- Optionally use the built-in preview to confirm the original plays back correctly and that the duration is within the 5-minute decoded window.
- Select the Reverse audio action and wait while every decoded channel sample is reversed locally in the current page. The number of frames and the channel count reported after decoding determine how long this step takes.
- Preview the reversed result and verify the duration, channel count, sample rate, and file size displayed by the tool so you can spot any unexpected change.
- Download the new file. It arrives as a freshly encoded PCM16 WAV with the reversed audio ready to move into an editor or share.
Selecting another file, running the operation a second time, or leaving the page releases the object URLs that backed the previous preview and download, so you do not need to clean up by hand. The Web Audio context used to decode the file is also closed after the operation ends.
What the Output WAV Contains and What It Drops
The download is always a freshly encoded PCM16 WAV, never the original container played backward. The browser-decoded floating-point samples are clipped to the range from -1 to 1, mapped to signed 16-bit PCM, and written little-endian so that negative full scale maps to -32768 and positive full scale maps to 32767. Reversing samples preserves the decoded duration, sample rate, and channel count when those values fit the stated limits. A stereo input stays stereo, with the left channel reversed independently from the right and both written back in the correct interleaved frame order. A WAV file can be considerably larger than a compressed MP3 of the same content, because every sample is now stored as a full 16-bit integer instead of relying on perceptual encoding. Reducing 32-bit floats to 16-bit integers also changes numerical precision, which matters if the source had headroom above full scale.
Metadata is not promised. Album art, ID3 tags, chapters, loop markers, encoder settings, the original compressed bitrate, and other container fields are not carried into the result. The tool reverses decoded audible sample data rather than editing the source container in place. Loudness is not normalized, silence is not trimmed, pitch is not shifted on purpose, time is not stretched, and clipping is not repaired.
| Property after reversal | What the WAV holds |
|---|---|
| Decoded duration | Preserved exactly |
| Sample rate | Preserved when inside 8,000 to 192,000 Hz |
| Channel count and stereo interleaving | Preserved (1 to 8 channels) |
| Container (MP3, M4A, AAC, Ogg, WebM) | Replaced with PCM16 WAV |
| Tags, album art, chapters, loop markers, bitrate | Not preserved |
| Original loudness, pitch, length, clipping | Not changed by the tool |
When Local Processing Helps and Where It Stops
Local processing is a clear advantage when the recording is sensitive — voice memos, internal drafts, or anything covered by a privacy expectation — or when the network will not cooperate. The tool also suits sound-design experiments, reverse speech, transition effects, sample inspection, and listening to the end of a recording from a different perspective. Those use cases are spelled out in the product contract and depend on the audio never leaving the device.
A local reverse pass does not cover every audio task. If you need to cut the reversed result to a specific time range, the Audio Cutter shares the same local PCM16 WAV workflow and is the natural follow-up step. If the result also needs a deterministic echo, reverb, fade, or normalization, Audio Effects layers those on the same file in the same tab. For a parallel question about whether the cutter uploads, see the parallel upload check for the Audio Cutter. For a more general walkthrough of reversing sound on a file without leaving your machine, see the local reverse-sound walkthrough.
The reversal logic and encoder are shared by the production tool and isolated tests. Reversal tests cover odd and even frame counts, multiple channels, and confirmation that source arrays are not mutated. PCM tests cover values below -1, above 1, negative and positive half scale, endpoints, and non-finite input. WAV golden tests inspect the RIFF, WAVE, fmt, and data fields, byte rate, block alignment, little-endian PCM bytes, and stereo interleaving. Those checks catch common failures such as reversing channel order instead of time, dropping the centre sample of an odd-length clip, or writing an invalid header. Listening to the original preview, generating the reversed WAV, and then listening to both the beginning and end of the result, while confirming the duration, channel count, sample rate, and file size the tool displays, is the most reliable way to confirm the output matches what you intended and that no upload took place.