Audio
Converting audio to MP3: which bitrate to choose, with real figures
MP3 is the most universal audio format: phones, car stereos, editing software and messaging apps all play it. Choosing it is rarely a mistake. The real question is the bitrate, expressed in kilobits per second (kbps), which determines both the file size and what the encoding sacrifices. To answer with figures rather than impressions, we converted two real recordings with FileXvert's settings.
Lossy and lossless: the distinction that matters
WAV stores sound exactly as it was digitised, sample by sample, with no compression at all. FLAC applies lossless compression, similar to a ZIP file: the decoded sound is strictly identical to the WAV, but the file is smaller. Both formats are meant for archiving and editing.
MP3, AAC (often in a .m4a file) and OGG Vorbis are lossy formats. They rely on a model of human hearing to remove what the ear perceives poorly: very high frequencies, and sounds masked by a louder sound at the same moment. The lower the bitrate, the more the encoder removes.
Measurement 1: an ambient recording
First file: a recording of a thunderstorm on a car windscreen, from the free sound library La Sonothèque. It is a 52.9-second mono WAV, 24-bit at 48 kHz, weighing 7.62 MB. Rain is a hard signal for an encoder: dense noise, rich in high frequencies, with no silence and no repetition.
| Format | Size | Share of the original WAV |
|---|---|---|
| WAV 24-bit (source) | 7.62 MB | 100% |
| FLAC (lossless) | 4.74 MB | 62% |
| WAV produced by FileXvert (16-bit) | 5.08 MB | 67% |
| MP3 192 kbps (no reduction) | 1.27 MB | 17% |
| MP3 128 kbps (balanced) | 848 KB | 11% |
| MP3 96 kbps (smallest) | 636 KB | 8% |
| AAC / M4A 192 kbps | 1.28 MB | 17% |
| OGG Vorbis 192 kbps | 1.51 MB | 20% |
FLAC saves 38% here without losing anything. That is less spectacular than with studio-recorded music, because rain noise is by nature unpredictable and therefore hard to compress. A 192 kbps MP3 brings the file down to 17% of its original size.
A useful detail: the WAV FileXvert produces is 16-bit, the resolution of audio CDs, which is more than enough for listening. If you work in a studio with 24-bit files, keep your originals: converting them to WAV with FileXvert would reduce their resolution.
Measurement 2: converting an MP3 to FLAC achieves nothing
Second file: a 3 min 28 s song, already an MP3 at 128 kbps, weighing 3.32 MB. It is the typical case of a track downloaded years ago.
| Conversion | Size | Relative to the source |
|---|---|---|
| MP3 128 kbps (source) | 3.32 MB | × 1 |
| MP3 192 kbps | 4.99 MB | × 1.5 |
| FLAC | 35.7 MB | × 10.7 |
| WAV | 36.6 MB | × 11 |
The resulting FLAC is almost eleven times larger than the MP3, yet it does not contain a single extra piece of information: it faithfully preserves the sound exactly as the MP3 encoder had already degraded it. The same goes for an MP3 re-encoded at 192 kbps: heavier, not better, and even very slightly degraded by this second pass through a lossy encoder.
Converting an MP3 to WAV does have a legitimate use, though: some editing software and older equipment only accept WAV. In that case the conversion solves a compatibility problem, not a quality one.
Which bitrate to choose in FileXvert?
FileXvert offers three levels for MP3, AAC and OGG: 192 kbps with no reduction, 128 kbps for balanced, and 96 kbps for smallest size. Here is how to use them:
- Speech (podcast, voice memo, lecture): 96 kbps is plenty. The voice occupies a narrow frequency band and tolerates compression well.
- Music for everyday listening, on the phone or in the car: 128 kbps is a reasonable compromise; 192 kbps if you listen on headphones and space is not an issue.
- Music to archive or edit: neither. Keep the WAV or convert to FLAC, then create MP3s from that original whenever you need them.
- A file for an Apple device: AAC in .m4a gives the same size as MP3 at an equal bitrate, and plays natively.
One simple rule avoids most disappointments: always convert from the best available source, and never run an already compressed file through a lossy encoder again if you can avoid it.
Checking the result
Listen to the converted file on headphones, with the earphones you normally use, rather than on laptop speakers, which hide most flaws. The passages that betray excessive compression are cymbals, applause, rain and sibilant voices: a watery or metallic sound means the bitrate is too low.
Also check the beginning and end of the file, and the total duration shown by your player, which should match the original.