Audio Compressor

Source audio:

WAV, MP3, M4A, AAC, OGG, FLAC, or WebM audio. One file at a time.

{{ sourceFile ? sourceFileSummary : 'Your source never leaves this device' }}

{{ processing ? `Compressing locally · ${processingProgress}%` : sourceStatus }}

The generated one-second WAV safely exercises compression without using your file.

Ignored {{ ignoredFileCount }} extra {{ ignoredFileCount === 1 ? 'file' : 'files' }}. This tool compresses one file at a time.

Target bitrate:

128 kbit/s is the balanced default. Choose 64 kbit/s for speech or 192–256 kbit/s when music detail matters more.

Compression:

Compression starts only after this action and runs locally.

Sample rate:

Auto is neutral: it resolves to 32, 44.1, or 48 kHz from the source. Pin a rate only for destination compatibility.

Channel handling:

Auto is neutral: mono stays mono, stereo stays stereo, and wider layouts are reduced to stereo.

Compressed audio

Review MP3 playback

Revision {{ outputRevision }} · {{ computation.values.target_bitrate }} kbit/s · {{ formatSampleRate(computation.values.output_sample_rate) }} · {{ channelLabel(computation.values.output_channels) }}

Verified

The chart renderer is unavailable. The same byte counts remain available in the compression ledger.

Checkpoint	Source	Output	Meaning	Copy
{{ row.label }}	{{ row.source }}	{{ row.output }}	{{ row.detail }}

Audio file size is shaped mainly by duration, bitrate, codec, and channel count. Uncompressed audio stores far more data than speech or music usually needs for sharing. Lossy compression removes detail that is less perceptible, then encodes the remaining signal at a target data rate. The result can be dramatically smaller, but each additional reduction narrows the margin for music detail, transients, ambience, and repeated editing.

Bitrate is the strongest direct size control for a constant-bitrate MP3. A 64 kbit/s target can suit spoken voice when compact delivery matters. A 128 kbit/s target is a balanced starting point. Music often benefits from 192 or 256 kbit/s, especially when cymbals, reverberation, stereo width, or dense arrangements need to survive the conversion.

Sample rate controls how many times per second the waveform is sampled. Lowering it does not guarantee proportional file savings when bitrate is already fixed, but it can affect frequency range and destination compatibility.
Channels describe mono, stereo, or wider layouts. Converting speech to mono can remove unnecessary stereo data; forcing music to mono removes spatial separation.
Transcoding decodes one compressed format and encodes it again. Quality already lost by the source cannot be restored, and another lossy pass can add artifacts.

A smaller byte count is only one acceptance test. The output should decode as MP3, keep the expected duration, use the intended sample rate and channel count, and sound acceptable on the devices or services that will receive it. Listening to a representative quiet passage and a dense passage usually reveals problems faster than the savings percentage alone.

Keep a lossless or highest-quality original when the recording may be edited again. A compressed delivery copy is useful for upload limits, previews, voice notes, and routine sharing; it should not become the only master simply because it is convenient.

How to Use This Tool:

Choose the quality tradeoff for the content first, then confirm the generated MP3 rather than relying on an estimate.

Choose one WAV, MP3, M4A, AAC, OGG, FLAC, or WebM audio file. The source must be audio-only, between 0.1 and 600 seconds long, and no larger than 25 MB.
Select Target bitrate. Start at 64 kbit/s for speech, 128 kbit/s for a balanced copy, or 192 to 256 kbit/s when music detail matters more than minimum size.
Leave Sample rate on Auto unless the destination requires 32, 44.1, or 48 kHz. Leave Channel handling on Auto to retain mono or stereo and reduce wider layouts to stereo; choose Mono or Stereo only for a deliberate conversion.
Choose Compress to MP3. Changing any setting after a completed run invalidates the previous result, so compress again before using the new configuration.
Play Compressed audio, compare the verified source and output sizes, and check the ledger for duration, bitrate, sample rate, channels, codec, and verification status before downloading.

Interpreting Results:

The displayed savings percentage uses the actual generated MP3 size after verification. Compression succeeds only when the output is readable, its duration stays within tolerance, its sample rate and channel count match the plan, and its byte count is lower than the source.

A large saving does not imply good sound. Listen for watery high frequencies, smeared transients, pumping, reduced stereo width, and speech consonants that become difficult to understand. If the output fails the listening check, increase the bitrate or keep the original format.

Source size can make the percentage look unusual. A highly compressed MP3 may not become smaller at an equal or higher target bitrate, while a WAV or FLAC source can show large savings. Compare output quality and bytes against the actual delivery requirement, not against a universal percentage target.

Technical Details:

Compression decodes the primary audio track, discards video tracks, resamples when necessary, applies the chosen channel count, and encodes a constant-bitrate MP3. Auto sample rate resolves to 48 kHz for sources at or above 48 kHz, 44.1 kHz for sources at or above 44.1 kHz, and 32 kHz for lower supported rates. Auto channels preserve one or two channels and reduce wider layouts to stereo.

Formula Core:

Before encoding, the expected MP3 payload size is estimated from duration and target bitrate.

B_{estimate} = \frac{t r 1000}{8}

t is duration in seconds, r is target bitrate in kbit/s, and B is bytes. The estimate is rounded up to a whole byte. A three-minute recording at 128 kbit/s has an estimated payload of 2,880,000 bytes before actual encoding settles the final container size.

After encoding, savings use measured source and output bytes.

Savings percent = \frac{B_{source} - B_{output}}{B_{source}} \times 100

Transformation Core:

Audio compression and verification stages
Stage	Requirement	Failure meaning
Read	Recognize the container and find one usable primary audio track with no video track.	The file is unsupported, damaged, empty, or not audio-only.
Plan	Resolve bitrate, sample rate, and one- or two-channel output.	A setting is outside the supported choices or the source properties are out of range.
Encode	Force a new MP3 transcode at 64, 96, 128, 192, or 256 kbit/s.	The browser cannot encode the track or layout.
Verify	Decode the result and confirm its track, sample rate, channels, duration, and size.	No download is offered for an unreadable, mismatched, or non-smaller result.

Duration may differ by no more than the larger of 120 milliseconds or 2% of source duration. Sample rate and channel count must match exactly. Those checks catch incomplete or misconfigured output, but they do not measure audible quality.

Privacy and Compatibility Notes:

The audio is decoded, encoded, and verified in the browser rather than uploaded to a media-processing service. Conversion support may still need to load over the network before the first run.

The output is always MP3, regardless of the source format.
Browser codec support and available memory can affect whether a valid source completes.
Tags and source metadata are not carried into the new file.
Auto channel handling reduces sources wider than stereo to two channels.
Keep the original until the MP3 has passed listening and destination-compatibility checks.

References:

Converting media files, Mediabunny documentation.