Somebody at a Bitcoin party last night told me a mutual fit 11 songs onto 4 megabytes.

Now, I'm not the type to argue at parties. But I'm absolutely the type to go home, open a terminal, and run the numbers until I feel better about the world.

So I did. And the math — as math tends to do — snitched immediately.

Here's the thing about audio files: they're not magic. They're math wrapped in a container. Every second of music you hear takes up a specific amount of space based on one key variable — bitrate. Bitrate is how many kilobits per second your file is pushing. The higher the number, the more information per second, the better it sounds, the more space it takes up. Simple.

So when somebody tells you 11 songs live happily inside 4 megabytes, your first instinct should be to pull out a napkin and start doing long division. Here's how it breaks down:

A standard song is roughly 3 minutes — 180 seconds. Eleven of those gives you 1,980 seconds of audio. Four megabytes converts to 32,768 kilobits total. Divide one by the other and you get ~16.5 kbps.

Sixteen. Kilobits. Per second.

For reference: the bottom floor of what most streaming platforms will even accept is 64 kbps, and that already sounds like music coming through a wall from your neighbor's apartment. At 128 kbps — considered "standard" MP3 quality — a single 3-minute song runs about 3MB. Meaning one decent-sounding song is nearly the size of the entire 4MB we're talking about.

So yes, you can technically fit 11 songs into 4MB. The same way you can technically fit into jeans from 2014. Possible. Not pretty. Nobody should have to witness it.

What 16 Kbps Actually Sounds Like

Audio engineers call it the "watery" effect. Producers call it "metallic shimmer." I call it sounding like someone recorded your music through a tin can and then ran it through a blender set to "anxiety."

At that bitrate, the bass is functionally gone. The stereo field collapses. High-frequency detail dissolves. What you're left with isn't really music — it's the idea of music, compressed to the point of suggestion. The codec is doing its absolute best to reconstruct what you gave it, but at 16 kbps, it's basically a psychic trying to describe a painting it's never seen.

The best codec in the world right now for this kind of extreme compression is Opus — and even Opus, the most efficient audio codec currently in existence, will humble you at 16 kbps. Opus was built for low-latency voice communication. It handles speech remarkably well at these sizes. Music, though? Music has dynamics, bass, spatial depth, harmonic complexity. Music asks more of the codec than the codec can give at that rate.

To be fair, if you're going to attempt this, Opus is your only real option. You'd convert everything to mono (stereo doubles the data requirement), drop the sample rate to 24 or 32 kHz, and run something like:

ffmpeg -i input.mp3 -c:a libopus -b:a 16k -ac 1 output.opus

And what comes out the other side is technically music. In the same way a photocopy of a photocopy is technically a document.

I know what you're thinking: so when would I ever need this? And the honest answer is — more often than you'd expect, just not always in the way you'd expect.

Compression isn't a failure mode. It's a tool. And like every tool, the key is knowing what job it's actually built for.

Game audio and interactive media. If you're scoring a mobile game, a browser-based experience, or a web app, you're fighting file size constantly. Background loops, ambient textures, sound effects — these live in highly compressed formats not because quality doesn't matter, but because load time and storage budgets are real constraints. A 30-second looping ambient track at 32 kbps still sounds fine as an atmospheric bed when a player is focused on a game. Knowing how to hit a 500KB ceiling without your music sounding like a dial-up modem is an actual skill with real demand.

Previews and streaming embeds. Platforms like SoundCloud's older player, some social media preview systems, and embedded widgets serve audio at compressed rates by default. If you want your music to translate across those environments, understanding what survives compression — and what doesn't — changes how you mix and master. Tracks with heavy low-end and lots of stereo spread suffer the most. Tracks with clear mid-range and controlled dynamics survive better. That's not an accident; that's craft.

Archival triage. If you have hundreds of rough sessions, scratch demos, or reference recordings and need to archive them affordably, strategic compression lets you retain the content — the performance, the melody, the arrangement — without paying full storage costs for material that may never get touched again. The idea is there. The vibe is there. The fidelity can come later when it matters.

Satellite radio and broadcast formats. Certain broadcast chains, especially international ones, transmit at lower bitrates due to bandwidth constraints. Knowing how your music behaves at 64 kbps before you pitch it to those channels is information that can change how you prepare your masters.

The Bitcoin inscription angle. I want to give this a proper mention because it's genuinely fascinating, even if the audio quality conversation leads you right back here. There's a growing movement of musicians inscribing audio onto Bitcoin — permanent, immutable, timestamped on-chain. The constraint is block size, which means you're working with very small files. At present, that means the audio quality tradeoffs we've been discussing throughout this entire piece are baked into the format. For certain kinds of sonic art — drone, texture, generative sound — that constraint might actually serve the work. But if you're trying to put a pristine mix on-chain and have it represent your best creative output? We're not there yet. The priority, as always, is the music. The medium catches up eventually.

You don't have to become a codec engineer. But you should know that bitrate is the variable, Opus is the best tool for low-bitrate survival, and anyone who tells you 11 songs fit into 4MB using compression with no sonic sacrifice is producing incredibly short songs, and in low fidelity, or is using magic.

Either way — now you know what I know.

The musician who understands the technical landscape of their work is the one who doesn't get surprised by it. Knowing that your 48kHz stereo master is 100MB+ and knowing exactly how to compress it to 3MB without losing what matters — that's leverage. That's the difference between an artist who is at the mercy of platforms and one who arrives prepared.

GIRL BARS by Endodeca — a blog on music creation, music publishing, music distribution, onchain music, immersive music experiences—creating them, and music technology.