Which MP3 bitrate? 128 vs 192 vs 320 kbps, decided
Use 192 kbps for music, 128 kbps for speech, and 320 kbps only when the MP3 is an archive copy or will be edited again. That one sentence settles most conversions; everything below is the file-size math, the reasoning, and the edge cases where the default answer bends.
The comparison
| Bitrate | Best for | Quality reality | Size per hour |
|---|---|---|---|
| 128 kbps | Podcasts, voice memos, lectures | Transparent for speech; music loses cymbal detail and reverb tails | 57.6 MB |
| 192 kbps | Everyday music listening | Indistinguishable from the source for most listeners on most gear | 86.4 MB |
| 256 kbps | Nervous perfectionists | Closes the gap on the rare dense tracks where 192 slips | 115.2 MB |
| 320 kbps | Archives, DJ sets, files that get re-edited | The best MP3 can do; beyond the limit of blind tests | 144 MB |
The sizes are not estimates. MP3 at a fixed bitrate is arithmetic: bits per second, times seconds, divided by eight. A 90-minute lecture at 128 kbps is 86.4 MB. The same lecture at 320 kbps is 216 MB and sounds exactly the same, because it’s speech and 128 already captured all of it.
What bitrate actually is
Bitrate is the encoder’s per-second spending budget. At 128 kbps it gets 128,000 bits for each second of audio and spends them on whatever your ear ranks first (the loud, the central, the melodic) while discarding what hides underneath. A bigger budget keeps more of the quiet detail: room reverb, cymbal shimmer, the decay of a piano note. The returns diminish hard, though. The first 128 kbps buy nearly everything; the last 128 buy things most people never hear.
Speech is not music
A voice recording is one source in a narrow frequency band with plenty of silence. That’s an easy job for an encoder, which is why 128 kbps is transparent for speech: podcasts, lectures, interviews, meeting recordings, all of it. Going higher buys megabytes and nothing else.
Two extra wins while you’re at it. A single voice carries no stereo information, so flipping the mono switch halves the file at no audible cost. And if the recording is headed for a transcription service, smaller is actively better. See preparing audio for transcription. The phone-recording workflow itself is covered in voice memos to MP3.
Music is the hard case: wide frequency range, many sources at once, dense passages. At 128 kbps a busy mix audibly smears: cymbals turn to static, reverb tails vanish, applause becomes mush. 192 kbps is where music stops sounding encoded for the large majority of listeners on the large majority of equipment. That’s why it’s the default here.
When 320 earns its megabytes
320 kbps is the ceiling of the format, and it has legitimate uses:
- Your only copy. If the source gets deleted after conversion, encode the keeper at maximum quality. Storage is cheap; regret is not.
- Headroom for re-encoding. Every lossy re-encode stacks loss on top of the last one. A file that will be trimmed, remixed, or converted again later should start as high as possible.
- Big playback systems. DJ sets and PA rigs push material through processing a phone never applies, and the extra detail survives it better.
And the cases where it’s waste:
- Phone, car, and desk listening. You pay 67% more storage than 192 for a difference that reliably fails blind tests.
- Anything that began life as a stream or a low-bitrate download. See the next section.
- “Just to be safe.” Safety is a lossless original, not a bigger MP3. Keep the FLAC and make MP3s from it for the devices that need them.
Converting up restores nothing
Take a 128 kbps file and convert it to 320, and you get a file 2.5 times larger that sounds exactly like 128. A hair worse, actually, since the re-encode adds its own small loss on top. The detail was discarded at the first encoding, and no setting reaches back in time to fetch it.
Transcoding is a photocopier, not a scanner.
The rule: match the source bitrate or go lower, never higher. The one honest reason to re-encode a 128 kbps file is compatibility (a format your device won’t play), and in that case 128 out is the right choice. If your real question is which format to keep long-term, MP3 vs AAC covers that fight.
How mp3fy handles bitrate
The converter offers 128, 192, 256 and 320 kbps with 192 as the default, plus a mono/stereo switch, and your tags and cover art carry over untouched. Conversion runs on-device in your browser at typically 20–80× realtime, so your files never upload no matter which bitrate you pick. And since the machine doing the work is yours, there are no size or duration limits either. The full model is on the privacy page.
One setting also covers a whole folder of files at once: pick 128 for a semester of lectures, convert, get a ZIP.
Frequently asked
Is 320 kbps overkill?
For listening, almost always. Controlled blind tests have shown for two decades that most people can’t tell 192 from 320 on normal equipment, and nearly nobody does it reliably. It stops being overkill the moment the MP3 becomes a master copy: the only version you’ll keep, or one that faces another round of editing and re-encoding.
Can I hear the difference between 192 and 320?
There’s a fast way to find out. Encode the same track at both rates, load them in any player, and switch without looking. If you can’t pick the 320 file at better than chance (and on most gear, with most music, you won’t), you’ve just saved 40% of your library’s storage. Trained listeners with studio monitors and known killer samples (harpsichord, applause, crash-heavy drums) sometimes pass, which mostly proves how contrived the winning conditions are.
What does mp3fy use by default?
192 kbps, with mono or stereo as its own switch. It has been the sensible middle ground since long before MP3’s patents expired in 2017, and it’s why mp3fy (converting files since 2015, rebuilt from scratch in 2026) ships it as the default: good enough that almost nobody should change it, small enough that nobody resents it.