Bit Rate, Sample Rate, and Why Your 64 kbps MP3 Sounds Like a Phone Call

Start here: file size is not a quality rating

Someone in a group chat asked the other day why a song they'd encoded at 64 kbps sounded like it was coming through a car phone. Someone else told them to just use APE instead, since it's lossless, and who cares if it's big.

One file squeezed past the point of recognition, one clinging to lossless. Neither answer was wrong, and neither one addressed the actual question.

Compression is a trade. You're deciding what to give up. If bit rate, sample rate and lossy-versus-lossless aren't straight in your head, you're picking formats by vibes — and vibes are how you end up with a folder full of regrets.

So let's go back to the beginning, then put the old contenders side by side: MP3, WMA, OGG, AAC, APE, FLAC. Old knowledge, still used every single day.

1. Two numbers you need before anything else

Bit rate: how much you feed it per second

You've seen the labels: 128 Kbps, 320 Kbps, 1411 Kbps. Kbps is kilobits per second.

One line: the bigger the number, the more audio information lands in each second, and the finer the result should be.

Red Book CD audio sits at 1411 Kbps. A 128 kbps MP3 carries roughly a twelfth of that. On a three-minute song, the encoder is reconstructing the original from one-tenth of the data, and what it drops is exactly the stuff it has decided your ears don't care about.

So don't be surprised that MP3 sounds soft. That's the deal. And if you genuinely can't hear the difference — no shame, truly. Enjoy your music 🫖

Sample rate: how many slices per second

Expressed as 44100 Hz, 32000 Hz and so on.

Picture the waveform as a continuous line. Sampling is putting a dot on it at fixed intervals. 44.1k means 44,100 dots a second; 32k means 32,000. More dots, closer to the original shape.

Nyquist's rule still holds: to reproduce a frequency of f, you need to sample at 2f or higher. Human hearing tops out around 20 kHz, so CD settled on 44.1 kHz with a little headroom. That number was calculated, not chosen because it looked nice.

Higher sample rate doesn't automatically mean better, though. Read it together with bit depth and bit rate, or you'll end up comparing numbers that don't mean what you think they mean.

2. Lossy: what the shrinking machines actually do

Lossy means the file you get back out isn't the file you put in. It cut what it judged you wouldn't notice. These are the ones that fought the format war.

MP3: the ancestor of all of them

MPEG Audio Layer III, built around 1993 by Fraunhofer IIS and Thomson in Germany. Compression of 1:10 or better — a 30 MB song down to three or four, roughly 1 MB per minute. That ratio is the entire reason it owned digital music for a decade.

The encoding modes matter:

• CBR (constant bit rate): the old default, everything at 128 kbps, no imagination required
• ABR (average): hits a target across the file
• VBR (variable): more bits for the busy passages, fewer for the quiet ones. This is the smart one

LAME is the reason modern MP3 still holds up — a good VBR encode at 320 kbps remains genuinely competitive.

The ceiling is real, though: 320 kbps is the top, and the high end gets sliced off. Don't mythologize it.

WMA: Microsoft showed up with DRM

Windows Media Audio. The pitch was aggressive compression without audible damage — Microsoft claimed around 1:18, half the size of the same thing as MP3.

The other hook was DRM: limit copies, expire playback, lock a file to one machine. That was aimed squarely at MP3's total lack of copy protection.

The rule of thumb from people who lived through it:

• Below 128 kbps, WMA wins — smaller and it holds together better
• Above 128 kbps, MP3 pulls ahead

So the old "128k WMA is about 192k MP3" line is a rough guide, not a law. Trust your ears.

OGG Vorbis: the open-source holdout

Careful: Ogg is the container, Vorbis is the codec inside it. Different things.

The selling point was free and unencumbered, no licensing fees, plus multichannel support — people genuinely hoped it would carry surround material. It runs up to 500 kbps and the algorithm is newer than MP3's, which puts it in the top tier of lossy quality.

The problem has always been support. MP3 plays anywhere; OGG depends on the player, and phones often needed an app. It still shows up in game audio and open-source projects, but on the web Opus and AAC have taken most of that ground.

MP3pro: great tech, killed by licensing

Thomson and Fraunhofer shipped it in 2001. Clever design: two layers, the bottom one a normal MP3, the top one using spectral band replication to restore detail. Same size, better sound; same sound, half the size.

The neat part was the extension — still .mp3. Old players played the base layer, new players decoded the full thing. Backward and forward compatible at once.

It didn't win because the licensing was expensive and the ecosystem never showed up. Good technology loses all the time in audio.

RA, ASF, VQF, MOD, MD: a museum wing

• RA (RealAudio): RealNetworks owned streaming back when 56k modems screamed. It played while it downloaded and dropped quality to match your bandwidth
• ASF: Microsoft's streaming container, same family as WMA
• VQF: Yamaha and NTT. Near-CD at 16-bit/44.1k, around 1:18, noticeably smaller than MP3. Closed spec and weak marketing, and that was that
• MOD: DOS-era game music. MIDI-like but built from real samples, with offspring like S3M, XM and IT
• MD: Sony's MiniDisc, 1992. ATRAC at 1:5, and editing that felt like magic — split, merge, rename, right on the device

You'll never need these. Knowing they existed keeps you from looking lost when an older engineer brings them up.

AAC and M4A: the ones actually in service

AAC (Advanced Audio Coding) came out of Fraunhofer, Dolby and AT&T under MPEG-2. The algorithm is simply newer: better sound per bit, up to 48 channels, multiple languages, 15 low-frequency effects tracks, and cheaper to decode on a battery.

It's the default answer on today's internet — YouTube, Apple Music, TikTok, iPhone ringtones (.m4r) are all AAC underneath. Common encoders are faac and Nero, topping out around 448 kbps.

M4A is just the container. Apple wanted to tell video MP4 apart from audio-only, so audio-only got .m4a. Inside is usually AAC, though it can hold Apple Lossless or even MP3 — same trick MKV pulls. Rename an audio-only .mp4 to .m4a and it will usually play, and vice versa.

MIDI: not audio, one more time

It stores a score and a set of instructions — a few KB for a whole piece, played back as synthesized instruments, no vocals. Teaching, arranging, old ringtones. Don't compare it for sound quality; there's nothing to compare.

3. Lossless: bigger, but byte-for-byte identical

Lossless means you get the original PCM/WAV back exactly. It removes redundancy, not sound.

CD and WAV: the raw baseline

Standard CD: 44.1 kHz / 16-bit / stereo, 1411 Kbps. Effectively the ceiling for "sounds like the source." WAV is Microsoft wrapping PCM in a RIFF header — the first four bytes literally read RIFF — and it sounds like the CD while being enormous and miserable to email.

And the classic trap, once more: CDA is not a file. Rip the disc with something like EAC into WAV or FLAC. Copying the icon gets you nothing.

AIFF (Apple) and AU (Unix) are in the same weight class, and editing software reads them all.

FLAC: open source won

Free Lossless Audio Codec, part of the Ogg family, free and open. Compression is roughly 2:1, which doesn't sound impressive until you look at the rest of the package:

• Decoding is integer-only, so even a cheap portable player can do it in real time
• Broader support than APE — most players just handle it
• Fast to encode, and it degrades gracefully when a file is damaged

The downside is that FLAC files run slightly larger than APE. Nobody cares about a few hundred megabytes when the alternative is a format you can't open.

APE: the old favorite

Monkey's Audio came first and was better known for a while. Bit rates in the 800–1400 kbps range, close to CD's 1411.2, at a bit over half the size of WAV, and bit-identical on decode. It did very well in blind listening tests back in the day.

But it encodes slowly, the compression edge over FLAC is small, and support is patchy. New archives go to FLAC, and APE is now mostly something you find in old downloads.

WavPack, Apple Lossless (ALAC), WMA Lossless and friends all have their niches. In practice FLAC plus ALAC covers essentially everything.

4. The rankings (enthusiast edition, with opinions attached)

Same source, same track, roughly speaking. Don't treat this as scripture.

Compression (smaller is more aggressive)

AAC ≈ OGG > MP3 ≈ WMA > APE > FLAC > WAV

Sound quality (lossless tied at the top)

WAV = FLAC = APE > AAC > OGG > MP3 > WMA

Combined quality + size + compatibility + encode speed (my call)

AAC > OGG > FLAC > APE > MP3 > WAV ≈ WMA

Situation	Pick this	Why
Sharing, messaging, uploading	320 kbps MP3 or 256 kbps AAC	Plays everywhere, size stays sane
Archiving, critical listening, masters	FLAC (ALAC if you're all Apple)	Recoverable later; not worth skimping
Ringtones, voice memos	AAC / m4a	Small and fine. Not WAV
Something you hand to a client	MP3 or AAC	WMA, VQF and APE will cause support tickets

The short version, free of charge:

• Sharing or uploading → 320 kbps MP3 or 256 kbps AAC
• Archiving or mastering → FLAC, or ALAC inside Apple's ecosystem
• Ringtones and voice memos → AAC or m4a, never WAV
• Don't deliver in WMA, VQF or APE — you'll be the one fixing it
• Lossless is not "twice as good." On modest gear you will not hear that last 10%

One last thing

Once this clicks, you stop asking why your MP3 sounds mushy, or whether keeping everything in APE makes you serious.

Compression is always a bit of a con — lossy cons your ears, lossless cons your hard drive. Know whether this file is going out or staying in, and the format picks itself.

When you need to actually convert something, run it through Audio Converter — common formats both ways, bit rate is yours to set — or pull the track out of a video with Video Converter. Everything runs in your browser: no uploads, no retention. Your privacy is fully protected.

Now if you'll excuse me, I have a few hundred gigabytes of APE to turn into FLAC. Great to listen to, exhausting to manage 😮💨

Common questions

Does a higher bit rate always mean better sound?

Within the same codec and the same source, basically yes: more bits per second means more audio information, so more detail. But it's one variable among several — sample rate, bit depth and the encoder all matter. A 64 kbps MP3 sounds like a phone call because the encoder was starved; 320 kbps is past what most ears can fault.

What's the real difference between lossy and lossless?

Lossy discards details it assumes you won't hear, so what comes back isn't what went in — MP3, AAC and OGG all work this way. Lossless removes redundancy instead, so decoding returns the original PCM/WAV byte for byte — FLAC, ALAC and APE. The cost is that lossless files are still several times larger.

Are M4A and AAC the same thing?

No. AAC is the codec, M4A is the container. Apple renamed audio-only MP4 files to .m4a to separate them from video, and what's inside is usually AAC, though it can be Apple Lossless or even MP3. Renaming an audio-only .mp4 to .m4a will typically play fine, and the reverse works too.

So what should I use day to day?

For sharing, messaging or uploading: 320 kbps MP3 or 256 kbps AAC. For archiving, critical listening or masters: FLAC, or ALAC on Apple gear. For ringtones and voice memos: AAC or m4a, never WAV. And don't hand anyone a WMA, VQF or APE — the compatibility problems become your problem.

Note: some of the material here was produced with AI assistance.

---

---