In this guide
Keep the original. If the complete recording must remain, test a lower-bitrate delivery copy and listen for lost detail. If only an excerpt is needed, shorten it deliberately. Compare the actual exported bytes with the destination's limit; a payload estimate alone does not prove that the file fits.
To make an audio file smaller, reduce the recording's duration or choose a delivery format and bitrate that use fewer bytes. Those changes solve different problems. Cutting an excerpt removes material; re-encoding to MP3 can discard audio detail. Keep an untouched original and check the exported file before sending it.
First confirm what the recipient accepts: file format, maximum file size, and whether the whole recording is needed. Do not assume that a smaller MP3 is suitable when the delivery specifically requires a PCM WAV. This guide uses a worked 1 MB budget, not a claim about any particular email or upload service's current limits.
Choose the change that fits the task
| Requirement | Useful first step | Tradeoff to check |
|---|---|---|
| Send the complete recording, and MP3 is accepted | Test an MP3 export at a lower bitrate | Lost detail, altered transients, or a file that is still too large |
| Send only a particular passage | Trim to the required excerpt | Missing words, attacks, tails, or the original timeline position |
| Keep PCM WAV for another edit | Check required duration and format settings before exporting | Changing bit depth, channels, or sample rate affects the working audio |
A level compressor, noise gate, or peak normalizer does not directly set a file-size budget. Those tools change dynamics or signal level. A PCM WAV with the same frame count, channel count, and bit depth keeps the same audio payload size even if every sample is made quieter.
Compress a short MP3 or WAV locally
- Choose the best source. Open the original in the free audio compressor. It supports mono or stereo MP3/WAV files up to 50 MiB and five minutes. It reads the selected file in your browser rather than uploading it.
- Select an MP3 bitrate. The choices are 64, 96, 128, and 192 kbps, with 128 kbps selected initially. A lower number uses fewer bits per second but may lose more detail. There is no rate that sounds appropriate for every recording.
- Enter your comparison limit. Choose MB or MiB to match the destination's stated unit. This field checks the output; it does not automatically choose a bitrate or guarantee acceptance by that destination.
- Encode and compare actual bytes. Select Encode MP3 & compare sizes. Read the original size, output size, and whether the actual export fits. The tool reports a larger or equal-sized export honestly instead of assuming a saving.
- Listen before downloading. Alternate between the original and the actual encoded MP3. Check quiet words, percussion, and sustained endings. Save a suitable result before leaving the page, and return to the original for another attempt.
Mono stays mono and stereo stays stereo. The browser decodes at 48 kHz; at low bitrates the encoder can choose a lower output sample rate. The MP3 is a new lossy encode, not a renamed WAV or a lossless copy. It can add delay or padding, and original tags or artwork are not copied.
Worked example: can you make the audio fit in 1 MB?
One decimal MB is 1,000,000 bytes. Divide that byte budget by duration to estimate the maximum encoded payload bitrate. The ceiling leaves no allowance for framing or other file data, so choose a rate below it and check the actual result.
Payload bitrate ceiling (kbps) = budget bytes × 8 ÷ seconds ÷ 1,000| Duration | Payload ceiling, rounded down | 64 kbps payload | 128 kbps payload |
|---|---|---|---|
| 1 minute | 133.33 kbps | 480,000 bytes | 960,000 bytes |
| 2 minutes | 66.66 kbps | 960,000 bytes | 1,920,000 bytes |
| 5 minutes | 26.66 kbps | 2,400,000 bytes | 4,800,000 bytes |
A one-minute recording at 128 kbps has a 960,000-byte payload estimate, leaving 40,000 bytes within that decimal budget for the rest of the file. A two-minute recording at 64 kbps has the same payload estimate. Neither calculation judges the listening quality or replaces checking the exported size.
For five minutes in 1 MB, the payload ceiling is only 26.66 kbps when rounded down. All four rates in the Saturalabs compressor exceed that budget. Lowering the comparison limit cannot make a 64 kbps export fit. Shorten the recording if an excerpt is acceptable, or choose another permitted delivery approach when every passage must remain.
One MiB is 1,048,576 bytes, so 1 MB and 1 MiB are different allowances. The audio file size calculator lets you compare those units, estimate maximum duration, and work backward from another budget. Leave headroom rather than choosing the theoretical maximum duration with no bytes left for overhead.
Shorten the recording when the excerpt is what matters
If the recipient needs one answer, scene, or sound, removing unrelated duration may be more useful than repeatedly lowering the bitrate of the entire recording. The free audio trimmer keeps one continuous time range from a supported short MP3 or WAV and offers WAV or re-encoded MP3 output.
Leave room around complete words, musical attacks, and decays. Keep a note of the source start time if the excerpt will return to a video or editing timeline. Trimming does not selectively remove an interruption from the middle or preserve every original timing relationship in a new lossy export.
For a longer file or unsupported format, use an editor that supports the complete source. Audacity's MP3 export documentation distinguishes constant, average, and variable bitrate modes. Use the appropriate export settings and inspect the actual file, rather than treating a quality preset as an exact byte-size prediction.
Why converting to mono or lowering volume may not help
For uncompressed PCM at the same duration, sample rate, and bit depth, one channel uses half the audio payload of two. But an MP3's selected total bitrate already covers its channels. Changing stereo to mono at the same total bitrate does not automatically halve its size.
Choose a channel because it contains the recording you need, not simply to make a number smaller. If an interview microphone was recorded only on one side, the channel splitter can extract that side as mono. It does not separate voices mixed on the same channel. Discarding a useful stereo side can lose another microphone, instrument, or spatial information.
Lowering volume changes sample values, not how many PCM sample frames are stored. Normalizing a WAV or fading its edges therefore does not make its PCM payload smaller. Keep size reduction separate from level adjustment and source cleanup.
Common questions
Can I make a WAV smaller without turning it into MP3?
You can shorten it or deliberately change PCM settings if the recipient permits that. Those changes do not preserve the same complete audio data. A suitable lossless format is another option when accepted, but the Saturalabs compressor exports MP3 rather than FLAC or a losslessly compressed WAV. Keep the original for editing.
Why did my compressed MP3 get larger?
Your source may already use a lower or variable bitrate, or be so short that framing matters. A new encode at a higher rate does not restore missing detail. Compare actual bytes and keep the original when a new export gives no useful storage benefit.
Will a file below an attachment limit always send?
No. A service can apply message-size rules, format restrictions, or other checks beyond the raw audio file's byte size. This guide does not prescribe a current limit for a particular service. Confirm the destination's rules and test the actual delivery.
Should I use the smaller copy for AI isolation?
Use the best supported source that already fits. If only a short test passage is needed, making that excerpt can avoid reducing the entire recording's bitrate. A more heavily compressed source does not promise better separation. The MP3 and WAV editing guide explains source, working-copy, and delivery choices.