Audio Compressor

Two kinds of “audio compressor”

An audio compressor can mean two things. This one makes files smaller: it writes the sound again as an MP3 that takes fewer bytes, so it is easier to email, upload or keep.

In music production, a compressor is an effect that narrows the gap between the loud and quiet parts of a sound, so a voice or a drum sits evenly in a mix. It changes how the sound behaves, not how much space the file takes. Making a file smaller doesn’t change its loudness range.

Which bitrate to choose when you compress MP3 files

The bitrate is how many kilobits each second of sound may use; the file’s size follows from it and the length. Lower bitrates leave out more of what the ear hears least, starting with the highest treble.

1. 1Speech, smallest (32 kbps, one channel): talks, lectures and voice notes where size matters most; clear, a little dull.
2. 2Speech (64 kbps, one channel): podcasts, interviews and audiobooks; sounds like the original to most listeners.
3. 3Music (128 kbps, two channels): songs to listen to on the go.
4. 4High quality (192 kbps, two channels): music you want to keep sounding full.

Listen to ten seconds at the setting you pick. A busy passage, such as applause or cymbals, shows the difference first.

How the exact size is known

At a constant bitrate every MP3 frame holds 1,152 samples (576 below 32 kHz) in a fixed number of bytes, give or take a padding byte now and then. The encoder adds a short delay before the sound and pads the end to whole frames. Add the frame that tells players about that delay, and the tags, and the size is fixed before anything is encoded.

This audio compressor uses one encoder in every browser and works its sizes out the same way; the files it writes match them to the byte. A variable bitrate would let busy passages take more, but then no size can be promised ahead, so this compressor writes constant bitrate.

One channel and the sample rate

Two channels carry stereo. When both carry the same sound, as in most voice recordings, one channel holds all of it, and the encoder can keep more of the treble at the same size: at 64 kbps one channel stays at 44.1 kHz, where two channels would drop to 24 kHz.

Lower bitrates sound best at lower sample rates, which carry less treble and leave more bits for the rest. On Auto, the audio compressor picks the rate the encoder itself would for each bitrate, such as 24 kHz at 64 kbps in stereo, and never one above the file’s own.

Compressing an MP3 again

An MP3 loses some detail each time it is encoded. Compressing one to a much lower bitrate is worth it; encoding it again at its own bitrate or higher only loses detail, which is why those settings are marked.

To reduce MP3 file size without losing anything, check what slimming saves first. A cover picture can be a tenth of a song’s file, and taking it out leaves the sound exactly as it was.

Size limits to aim for

Many email services stop at about 20 to 25 MB a message, and an attachment grows by about a third on the way, so a file well under the limit travels best. Chat apps and upload forms often have lower limits.

In size mode the audio compressor counts in megabytes of a million bytes, the stricter way, so a file shown under 25 MB is under it whichever way a service counts.