How it works
Your browser reads the file and decodes its audio track to raw samples. Those samples are then written again as MP3 or WAV by an encoder that runs in a background thread on your device, so the page stays responsive and you see the progress of each file. Files are converted one after another to keep memory use low.
The output is always 44.1 kHz, the sampling rate of an audio CD, at 16 bits. Audio recorded at another rate, such as the 48 kHz used by most video, is resampled. Surround sound is mixed down to stereo.
MP3 or WAV?
| MP3 | WAV | |
|---|---|---|
| Compression | Lossy: removes detail that is hard to hear | None: every sample is stored |
| Size of one minute in stereo | 1.4 MB at 192 kbps | 10.6 MB |
| Plays on | Practically everything | Practically everything |
| Use it for | Listening, sharing, phones, podcasts | Editing, music production, devices that require it |
How large will the file be?
An MP3’s size depends only on its length and bitrate, not on what it contains:
size in MB ≈ minutes × kbps × 0.0075
| Bitrate | One minute | A 4-minute song | A 1-hour recording |
|---|---|---|---|
| 128 kbps | 0.96 MB | 3.8 MB | 57.6 MB |
| 192 kbps | 1.44 MB | 5.8 MB | 86.4 MB |
| 256 kbps | 1.92 MB | 7.7 MB | 115.2 MB |
| 320 kbps | 2.40 MB | 9.6 MB | 144.0 MB |
A stereo WAV file takes 10.6 MB per minute, so the same 4-minute song is about 42 MB. Choosing mono halves a WAV file; for MP3 the size stays the same, but all of the bitrate goes to one channel, so speech sounds cleaner at 128 kbps.
Getting the best quality
- Convert from the original. Each conversion to MP3 loses a little. Going from WAV or FLAC to MP3 once is fine; going MP3 → MP3 at a lower bitrate adds the losses together.
- Do not raise the bitrate of an MP3. Converting a 128 kbps file to 320 kbps makes it larger, not better.
- Use mono for voice. A lecture or interview recorded on one microphone carries no stereo information worth keeping.
The MP3 encoder is lamejs, a JavaScript port of LAME, used under the LGPL licence.