MP3 bitrate guide: 128, 192, or 320 kbps?
Bitrate is the only quality control MP3 really gives you. Here is what each setting costs, what it sounds like, and when the difference stops mattering.
Quick answer. Music you care about: 320 kbps. General listening: 192 kbps. Speech, podcasts, and lectures: 96 to 128 kbps mono. Never re-encode an existing MP3 to a higher bitrate, because that only adds damage.
What bitrate actually controls
Bitrate is how many bits of data the encoder may spend on each second of audio. At 128 kbps it gets 128,000 bits per second, at 320 kbps two and a half times as many.
MP3 works by discarding audio it predicts you will not notice: sounds masked by louder ones, frequencies at the top of the range, detail in dense passages. A higher bitrate means the encoder has to throw away less to hit its target, so more of the original survives. That is the whole mechanism. There is no separate quality setting hiding somewhere.
The numbers side by side
| Bitrate | 1 minute | 4 minute track | 1 hour | Suits |
|---|---|---|---|---|
| 320 kbps | 2.3 MB | 9.2 MB | 137 MB | Music, deliverables, archives |
| 256 kbps | 1.8 MB | 7.3 MB | 110 MB | Music with a size budget |
| 192 kbps | 1.4 MB | 5.5 MB | 82 MB | Everyday listening |
| 160 kbps | 1.1 MB | 4.6 MB | 69 MB | Mixed speech and music |
| 128 kbps | 0.9 MB | 3.7 MB | 55 MB | Background music, long files |
| 96 kbps | 0.7 MB | 2.7 MB | 41 MB | Speech and podcasts |
| WAV, for comparison | 10.1 MB | 40.4 MB | 606 MB | Editing and mastering |
What each one sounds like
Descriptions of audio quality are slippery, so here is what to listen for rather than adjectives about warmth.
96 kbps. Fine for a voice. On music, cymbals turn watery and the stereo image narrows noticeably. You will hear it without trying.
128 kbps. The old default, and audible on demanding material. Applause becomes a wash of static, hi-hats develop a metallic shimmer that was not in the recording, and reverb tails end abruptly. On phone speakers, in a car, or under conversation, most of that disappears.
192 kbps. The point where problems get hard to find on ordinary equipment. Some listeners still catch it on solo piano or close-miked acoustic guitar, where there is nothing to hide behind.
256 and 320 kbps. Reliably identifying these against the original in a blind test is difficult, and plenty of confident listeners fail. The differences that remain live in the very top of the frequency range and in extremely quiet passages.
Worth being straight about: a great deal of what people describe as hearing bitrate turns out to be knowing which file is which. That does not make the data loss imaginary, it just means the audible stakes are lower than the arguments suggest.
Constant, variable, and average bitrate
Constant bitrate spends the same budget on every second, including silence. It is predictable, which is why some broadcast and telephony systems require it, and slightly wasteful.
Variable bitrate lets the encoder spend more on a dense chorus and less on a quiet intro, targeting a quality level instead of a size. LAME's presets run from V0, the highest, down to V9. V0 usually averages somewhere around 220 to 260 kbps on music while sounding comparable to 320 kbps constant bitrate. The exact average moves with the material, so a sparse recording lands lower.
Average bitrate sits between the two, varying within limits to hit a target size. It is rarely the right choice now that VBR is universally supported.
The converter here uses constant bitrate. That keeps behaviour identical across every browser and makes the file size predictable before you start, which matters when you are converting to fit an upload limit.
Three mistakes worth avoiding
Re-encoding an MP3 at a higher bitrate. Taking a 128 kbps file to 320 kbps gives you a file two and a half times larger containing the original damage plus fresh damage from the second pass. If 128 kbps is all you have, keep it.
Using 320 kbps for speech. A spoken word recording has far less going on than music, so the encoder hits transparency early. You are tripling the download size of your podcast for something nobody can hear.
Encoding from a WAV that came from an MP3. The WAV looks lossless and is not. Converting it back to MP3 is a second generation encode. Why that matters.
When MP3 is not the right format
AAC generally sounds better than MP3 at the same bitrate and is what Apple Music, YouTube, and most streaming platforms use. Opus is better still at low bitrates and is excellent for speech. MP3 survives because everything on earth plays it, which is a real advantage when you do not control the listener's device.
If the goal is keeping quality rather than saving space, skip lossy formats. FLAC is lossless, around half the size of WAV, and handles tags properly. The format comparison covers this.
Common questions
What is the best MP3 bitrate?
320 kbps if you want the most an MP3 can hold, 192 kbps if you want a sensible balance, and 96 to 128 kbps for speech. There is no single best number, because you are trading file size against how much audio the encoder is allowed to discard.
Can you hear the difference between 128 and 320 kbps?
On the right material, often yes. Cymbals, applause, and solo piano expose 128 kbps first. On a dense pop mix played through phone speakers or a car stereo, most people cannot. In a proper blind test the gap narrows a lot more than people expect.
Is 320 kbps the same as lossless?
No. 320 kbps is the highest bitrate MP3 supports, but it is still lossy: audio was removed to get there. Lossless means FLAC, ALAC, or WAV, where every original sample survives.
Does re-encoding an MP3 at a higher bitrate improve it?
No. Converting a 128 kbps MP3 to 320 kbps produces a larger file containing the same damaged audio, plus a second round of encoding damage on top. Always encode from the original WAV or FLAC where one exists.
What bitrate should a podcast use?
96 to 128 kbps mono is plenty for speech and keeps downloads small. Music beds and heavily produced shows benefit from 128 to 192 kbps stereo. Going higher mostly costs your listeners bandwidth for something they will not hear.
What does V0 mean?
It is LAME's highest variable bitrate preset. Rather than fixing the bitrate, the encoder targets a quality level and spends whatever data each passage needs, usually averaging somewhere around 220 to 260 kbps on music. It generally sounds as good as 320 kbps constant bitrate in a noticeably smaller file.