MP3 Bitrate Guide: 128 vs 256 vs 320 kbps (Which Should You Use?)

“What bitrate should I export this MP3 at?” is one of those questions where every answer online seems to contradict the last. One guide swears 320 kbps is the only acceptable setting; another says you’re wasting space above 128. Both can be right — because the correct bitrate depends entirely on what the file is for. A spoken-word podcast and a mastered orchestral recording have completely different needs, and matching the bitrate to the use case is how you avoid either bloated files or audible damage. This guide gives you a use-case-first recommendation, the file-size math behind each choice, and an honest read on whether the differences are even audible.

Quick answer: Speech, podcasts, audiobooks, voice memos → ~128 kbps. General music listening (phones, earbuds, car, streaming) → ~256 kbps. Archiving, mastering, critical listening, or files you’ll re-encode → 320 kbps. File size scales linearly: 128 kbps ≈ 1 MB/min, 256 kbps ≈ 1.9 MB/min, 320 kbps = 2.4 MB/min. Above ~256 kbps, the difference from lossless is hard for most people to hear in a blind test — so 320 buys headroom and peace of mind more than guaranteed audible improvement.

Jump to a section

• How bitrate maps to file size (the math)
• 128 kbps — speech, podcasts, voice
• 256 kbps — general music listening
• 320 kbps — archive, mastering, critical listening
• Is the difference actually audible?
• CBR vs VBR — a quick note
• Quick reference table
• Set a target bitrate with xconvert
• FAQ

How bitrate maps to file size (the math)

Bitrate is the single biggest driver of an MP3’s file size, and the relationship is simple linear arithmetic. Bitrate is quoted in kilobits per second (kbps), and there are 8 bits in a byte, so:

Worked out for the three common bitrates:

So a one-hour podcast at 128 kbps lands near 58 MB; the same hour at 320 kbps is roughly 144 MB — about 2.5× larger for content where the extra data does almost nothing (more on that below). The figures above use decimal megabytes (1 MB = 1,000,000 bytes), which is how audio tools and file managers usually report audio sizes. They’re also slightly idealized: the actual file is a little larger because ID3 tags (title, artist, cover art) add a small fixed overhead, and variable bitrate (VBR) files vary moment to moment around their average. For constant bitrate (CBR) files the math above is accurate to within the tag overhead.

If you want the deeper background on why bitrate and sample rate set quality and size — what those numbers physically represent — read our companion explainer, Audio Bitrate vs Sample Rate. This article is the decision guide: which number to actually pick.

128 kbps — speech, podcasts, voice

Use it for: podcasts, audiobooks, interviews, lectures, voice memos, voicemail-style recordings, and anything that is primarily spoken word.

The human voice occupies a relatively narrow frequency band and has little of the dense, wide-spectrum complexity that challenges an encoder. At 128 kbps, speech is clean, intelligible, and — for most listeners on most playback gear — indistinguishable from a higher bitrate. You get roughly 1 MB per minute, which keeps a long podcast episode small enough to download quickly on mobile data and stream without buffering.

Going above 128 kbps for pure speech is usually wasted space. The exception is speech with significant musical content — a podcast with full-song intros and stings, or an audiobook with a scored soundtrack — where bumping to 192 or 256 kbps protects the music. For talk-only content, 128 kbps (or even 96 kbps mono for a single speaker) is the efficient, well-matched choice.

256 kbps — general music listening

Use it for: music you’ll listen to on phones, earbuds, laptops, Bluetooth speakers, and car stereos — i.e. the vast majority of real-world music listening.

256 kbps is the practical sweet spot for music. It’s the bitrate at which, in blind listening tests, many people stop being able to reliably tell the MP3 apart from the lossless original (see the next section), while still being meaningfully smaller than 320 kbps. It’s also the bitrate several major download stores standardized on for AAC, which tells you something about where the industry put the quality/size line for everyday listening.

At ~1.9 MB per minute, a typical 4-minute track is around 7–8 MB — comfortable for a large on-device library without the storage cost of maxing out every file. If you rip or convert a music collection and you’re not sure you’ll ever need more, 256 kbps is the recommendation that ages well.

320 kbps — archive, mastering, critical listening

Use it for: master/archive copies, files you might re-encode or edit later, critical listening on high-end gear, and any situation where you want the maximum the MP3 format offers and don’t want to second-guess it.

320 kbps is the highest standard MP3 bitrate. Two genuinely good reasons to choose it:

• Re-encoding headroom. Every lossy re-encode compounds artifacts. If a file might be edited, trimmed, normalized, and exported again, starting from 320 kbps preserves more of the original signal through that chain. (Better still: keep a true lossless master like FLAC or WAV for archiving, and treat MP3 as a delivery format.)
• Peace of mind on demanding material. Dense, transient-rich music — cymbals, massed strings, applause, electronica with sharp attacks — is the hardest for an encoder. 320 kbps gives those passages the most bits to work with, so if anything is going to reveal a difference, this is where you want the headroom.

What 320 kbps is not is a guaranteed, audible upgrade over 256 for casual listening. For most people on most gear playing most music, it mainly buys a larger file (2.4 MB/min, ~144 MB/hour) and the confidence of knowing you didn’t cut a corner.

Is the difference actually audible?

Here’s the honest version, because this is where a lot of bitrate guides overpromise.

Below ~192 kbps, trained listeners can fairly reliably pick the MP3 out of a blind ABX comparison against lossless, especially on sharp, transient sounds. From roughly 256 kbps upward, the picture changes: in published blind tests, listeners’ ability to distinguish a good MP3 from the lossless source drops toward chance level. In one widely-cited blind test using the LAME encoder, the author could reliably identify the MP3 at 192 kbps but scored at chance (results like 7/10 and 6/10) at 256 kbps, concluding that 256 was the lowest bitrate they could not reliably tell from the original.

The important caveats:

• It’s contested and individual. “Audibly transparent” is not a fixed number. It depends on the specific track, the encoder and its settings, the playback equipment, the room, and the listener’s training. Some people on some material report hearing differences higher up; many cannot distinguish 256 from lossless at all.
• Real-world conditions narrow it further. On earbuds, in a car, or over Bluetooth, the playback chain masks far more than a quiet room with studio monitors reveals.
• Modern encoders are good. A well-tuned VBR encode targeting ~245 kbps average can be effectively transparent to most listeners — bitrate alone doesn’t tell the whole story.

So the practical takeaway is not “320 always sounds better.” It’s: 128 for speech, 256 as the safe everyday-music default, and 320 when you want maximum headroom or are archiving/re-encoding — with the understanding that the jump from 256 to 320 is often about insurance rather than an improvement you’ll consistently hear.

CBR vs VBR — a quick note

When you set a bitrate you can usually pick Constant Bitrate (CBR) or Variable Bitrate (VBR). CBR uses the same bitrate for every second of audio — predictable file size, simplest compatibility. VBR spends more bits on complex passages and fewer on simple ones, which generally gives better quality per megabyte. For music, VBR targeting a high average is an excellent choice; for streaming with strict bandwidth or maximum-compatibility needs, CBR is the safer pick. xconvert’s MP3 compressor lets you choose between Constant Bitrate and Variable Bitrate under its custom-bitrate mode.

Quick reference table

Bookmark this — matching bitrate to use case is the whole game; there is no single “best” number.

Set a target bitrate with xconvert

To change an MP3’s bitrate — drop a music export to 256 kbps, shrink a podcast to 128 kbps, or set a master at 320 kbps — use the xconvert MP3 compressor. Under Advanced Options, choose Custom Bitrate, pick Constant Bitrate or Variable Bitrate, and set your value. The same tool also offers Specific file size (target an exact size in MB) and File Size Percentage modes if you’re optimizing to a hard size limit rather than a bitrate.

Related xconvert tools and reading:

• Audio Compressor — the same controls for MP3, AAC, WAV, FLAC, OGG, and M4A when you’re not working with MP3 specifically.
• Audio Bitrate vs Sample Rate — the concept explainer behind this decision guide: what bitrate and sample rate actually measure.

Your files upload over an encrypted connection, are processed on our servers, and are deleted automatically a few hours later.

FAQ

Is 320 kbps always better than 256 kbps?

Not audibly, for most people. 320 kbps contains more data, but in blind listening tests the ability to distinguish a good 256 kbps MP3 from lossless already drops toward chance for many listeners — so the additional bits in 320 often don’t translate into a difference you’ll reliably hear on typical gear. 320 kbps is genuinely worth it for archiving, re-encoding headroom, and demanding material on high-end equipment. For everyday listening, 256 kbps is the practical default.

What bitrate should I use for a podcast?

128 kbps is the standard recommendation for spoken-word podcasts and audiobooks. The voice is clean and intelligible at that bitrate, and you get roughly 1 MB per minute, keeping episodes small for mobile downloads. If your show includes full-length music segments, step up to 192 kbps to protect the music; for a single speaker with no music, even 96 kbps mono can be fine.

How big will my MP3 be at each bitrate?

File size scales linearly with bitrate. Per minute: about 0.96 MB at 128 kbps, about 1.9 MB at 256 kbps, and 2.4 MB at 320 kbps. Per hour that’s roughly 58 MB, 115 MB, and 144 MB respectively. Add a small amount for ID3 tags and cover art, and note that VBR files vary around their average rather than hitting the figure exactly.

Does increasing the bitrate of an existing MP3 improve its quality?

No. Re-encoding a 128 kbps MP3 to 320 kbps does not recover detail that was discarded at 128 — the lost information is gone, and the new file is just a larger container for the same degraded audio. Set the bitrate you want at the point you encode from a lossless source. If all you have is a low-bitrate MP3, converting it up only wastes space.

Should I pick CBR or VBR?

For music, VBR (variable bitrate) targeting a high average generally gives better quality per megabyte, because it spends bits where the audio is complex and saves them where it isn’t. CBR (constant bitrate) gives a predictable file size and maximum compatibility, which can matter for some streaming or hardware playback scenarios. Both are available in xconvert’s custom-bitrate mode.

What’s the lowest bitrate I can use without it sounding bad?

It depends on the content. For pure speech, 96–128 kbps is typically clean. For music, audible artifacts become easier to detect below about 192 kbps, especially on sharp, transient sounds. There’s no universal floor — it varies by track, encoder, and listener — so when in doubt for music, 256 kbps is the safe choice.

Sources

Last verified 2026-06-18.

• AudioMountain — Audio File Size Calculations — bytes-per-second and per-minute/per-hour figures for 128 and 320 kbps MP3.
• Tom Szynalski — Blind-testing MP3 compression — LAME-encoder ABX results at 128/192/256 kbps and the transparency discussion (presented here as one cited, contested data point, not a universal claim).
• Colin Crawley — Audio File Size Calculator — cross-check for the bitrate-to-file-size arithmetic.
• xconvert — Audio Bitrate vs Sample Rate — companion concept explainer.