Compress audio to a target size in MB, or choose a quality level. Convert up to 20 MP3, WAV, M4A or other audio files to smaller MP3s and download them individually or as a ZIP.
Compare public-domain English narration with an original instrumental melody. These are real Toolkitta exports measured on 29 October 2026, not promised savings for every file.
Speech source and public-domain dedication
Reproduce the speech result: download the original WAV, add it above, choose Target size, enter 0.1 MB and tick Voice recording. The selected bitrate is 64 kbps. The result is 96,495 bytes, below the 100,000-byte target.
Reproduce the music result: add the original music WAV, choose Balanced (128 kbps) and leave Voice recording off. The result is 192,835 bytes. Listen for changes in the melody and at note endings before choosing a smaller level.
Each source is 12 seconds at 44.1 kHz. MP3 encoding adds a small amount of padding. These simple samples help compare the controls; test your own speech or music before sharing. The narration WAV is decoded from a public-domain MP3; it does not regain detail lost in that recording.
Enter the limit per file in MB, not the ZIP size. A 10-minute recording at 64 kbps is about 4.8 MB before headers; a 5 MB target may be a useful starting point. Check the exported size and leave room for other attachments. If the minimum bitrate cannot fit the whole recording, trim it or raise the target.
MP3 removes detail. WAV-to-MP3 usually saves much more than recompressing an already small MP3. Mono can improve speech at a low bitrate, but does not halve the size at the same bitrate. Keep the original and compare with headphones.
Trim a long recording first
An MP3 at 128 kbps uses about 0.96 MB per minute, 96 kbps about 0.72 MB and 64 kbps about 0.48 MB. A WAV from a CD-quality source uses about 10.6 MB per minute, so it shrinks a lot; an MP3 that is already 128 kbps barely changes at Balanced.
Toolkitta picks the highest standard MP3 bitrate whose file should fit your limit, leaving a small margin for headers and cover art, and caps it at 192 kbps. If a recording is too long to fit even at 32 kbps, you are told before compressing; trim it or raise the limit. The budget leaves 4% for overhead, so large cover art or tags can still push an output over the target. Check the result’s actual size.
Speech is usually recorded from one microphone. Mono spends the whole bitrate on one channel, so voices stay clearer at small sizes. The file is not smaller at the same bitrate, but a lower level sounds better.
No. Compression runs in your browser. Encoder code is downloaded from this site the first time it is needed; your audio never leaves your device.
Guides, common conversions and useful next steps.
Arrange clips, mix tracks and add fades on a local timeline.
Use the action beside a finished result to send it directly, or open another tool: