Compress Audio to a Target Size Without Uploading It

How it works

1. Your browser reads the audio from your disk. No network request carries the file.
2. It reads the duration and computes a bitrate from the target: target bytes × 8 ÷ seconds. It starts at the highest standard MP3 bitrate the maths says will fit.
3. It re-encodes the whole file to MP3 once. If the encoder overshoots, it steps down one bitrate rung and tries again. The result card shows the final bitrate, mono or stereo, and the length.
4. If the target is below MP3's floor, it tells you. It shows the smallest honest result it could produce instead of pretending it hit the number.

What it can and cannot do

The target is an upper bound, not an exact size. The tool stops at the first bitrate that fits, so the file is at or a little below the number you chose. We count 1 MB as 1,000,000 bytes, which is the stricter reading. The output is always MP3, at a standard constant bitrate from 8 to 320 kbps.

Low targets mean lower quality. 16–32 kbps can keep speech intelligible; music at that bitrate sounds rough. At 48 kbps and below the tool downmixes to mono, and the lowest bitrates also use a lower sample rate — both are better than splitting a tiny budget across two channels or a wide band. The result card shows what it used, so you can judge before downloading.

Metadata and cover art are not copied. The output is a fresh MP3 built from the decoded audio, so ID3 title, artist, album and embedded cover art are gone. A large cover image would also count against the target; dropping it is what keeps the size predictable. Tag the result afterwards if you need those fields.

Input formats depend on the browser's decoder. MP3, M4A/AAC, WAV, OGG/Opus and FLAC usually work in current Chrome and Edge. DRM-protected files and some uncommon codecs will not decode. This tool takes audio files, not video: if you drop a video, export its audio track first. Output is MP3 only; it does not produce M4A, AAC, OGG or WAV.

Free handles up to 100 MB and 30 minutes; Pro up to 300 MB and 2 hours. Both limits apply. An uncompressed WAV can hit the byte cap before the time cap — a 30-minute stereo WAV is about 318 MB — so export it as MP3 or M4A first if the page says it is too large.

It never returns a file larger than the one you gave it. If re-encoding cannot beat the original, it says so and leaves the original alone. It does not trim, join or edit the recording; it re-encodes the whole file.

The MP3 encoder is LAME, compiled to WebAssembly. It runs in this tab through @mediabunny/mp3-encoder (MPL-2.0). LAME itself is available under the LGPL. Nothing about the file is sent to either project.

Pricing

Questions

Is my audio really not uploaded?

Correct. Open developer tools, go to the Network tab, and compress a file. You will see the page, the script bundle and a few tiny counter pings — no request carrying the audio.

Will it hit exactly 16 MB?

It aims under, not exactly at. It keeps the highest standard bitrate whose file fits, so the download is at or below the target. That is what an upload form checks, and it avoids throwing away quality the limit does not require.

What bitrate will I get?

It depends on the length: bitrate ≈ target bytes × 8 ÷ seconds. A 30-minute recording at 16 MB has room for about 71 kbps, so the tool picks 64 kbps. The same recording at 5 MB has room for about 22 kbps, so it picks 16 kbps mono. The result card shows the exact choice.

Why is the result mono?

At 48 kbps and below, one channel gets the whole bit budget. Splitting a tiny budget across two channels usually sounds worse than a clean mono file. At 56 kbps and above, a stereo source stays stereo; a mono source always stays mono.

Will it sound bad?

At 16–32 kbps speech is usually intelligible, while music loses cymbals, reverb and detail. If the first result is too rough, choose a larger target, shorten the clip, or keep the original. The result card's bitrate is the number to compare.

Does it keep ID3 tags or cover art?

No. The output is a new MP3 built from the audio samples, so title, artist, album, comments and embedded cover art are not carried over. This also keeps a large cover image from pushing the file over the target. Tag the download afterwards if you need those fields.

Which formats can I use?

MP3, M4A/AAC, WAV, OGG/Opus and FLAC are accepted. What actually opens depends on the browser's decoder; current Chrome and Edge are the safest choice. DRM-protected files and some uncommon codecs will not decode. Output is always MP3.

Can I compress the audio from a video?

Not directly. This tool takes audio files; a video's audio track has to be exported first in a video editor or converter. Dropping an MP4 normally shows the "that is a video" message.

Will 16 MB fit WhatsApp or Discord?

If the app's current limit is 16 MB, choose 16 MB; the result will be at or under it. For a 10 MB limit, choose 10 MB. We do not control those limits and they change, so check the app's current help page before uploading a long file.

Can I do several files at once?

Free compresses one file at a time. Pro accepts up to 10 files in one drop and processes them one after another in the same tab.

What about a two-hour podcast?

Pro accepts up to two hours. A two-hour file at 16 MB only has room for about 18 kbps, so the tool uses 16 kbps mono; speech may be acceptable, music will not be. A 5 MB target is below MP3's 8 kbps floor at that length, so the page will show the smallest honest result instead.

Why MP3 and not M4A?

MP3 plays almost everywhere and the encoder is bundled, so the tool works without a server. M4A/AAC output is not available yet; use another tool if that format is required.

Do you store anything about me?

We count anonymous events — page opened, compression finished, upgrade clicked — with no advertising cookies, persistent analytics identifiers or audio data. An event may include the duration when a Free file exceeds the time limit. A strictly necessary signed cookie is used only after purchase so Pro access survives a refresh.