There Is No MP3 on YouTube: Why I Label the Audio Button M4A
When I first wired YouTube support into my downloader, the audio button said "MP3". Everybody expects an MP3 button. Every downloader site has one. So I added one without thinking.
Then I looked at what was actually coming back from the provider API.
What YouTube actually serves
YouTube does not ship MP3 files. The audio tracks inside YouTube's streams are M4A (AAC) or Opus in a WebM container. That is what the site itself streams to your browser. An MP3 file only exists if somebody decodes that audio and re-encodes it into MP3 — a conversion step with its own CPU cost, its own quality loss, and on shared hosting, a whole dependency chain (ffmpeg) I do not have and cannot install.
So the honest options are:
- Serve the audio exactly as YouTube provides it — M4A or Opus.
- Transcode to MP3 on the server — slow, lossy, and fragile on cheap hosting.
I chose option one.
Why transcoding on the fly is a bad deal on shared hosting
My site runs on plain shared hosting: PHP, no shell ffmpeg, no background workers, no Redis. Transcoding means spawning an ffmpeg process per request, holding a connection open while it works, and praying the host's process killer does not notice. One busy afternoon and the whole site crawls — or the host suspends the account.
Even if the plumbing were free, transcoding is lossy-to-lossy. YouTube's AAC at 128 kbps converted to MP3 at 128 kbps loses detail twice for zero benefit. The visitor gets a worse file, slower.
What the button says now
The audio button on ramrom.io says "Audio (M4A)". The homepage copy that used to promise MP3 now says M4A. It felt like a downgrade the day I made the change — MP3 is the word people search for — but the file a visitor receives is exactly what YouTube's own servers produced, and it plays everywhere: phones, browsers, car stereos. Nobody has complained. A couple of people noticed and said the download was fast, which is the part I actually care about.
The wider lesson
This is a small thing, but it is the same instinct behind every honest decision on the site: label what is actually there. A "4K MP4 with sound" button that hands you a 360p file with sound is the same lie as an "MP3" button that hands you M4A — it just sounds better in the marketing copy. I would rather ship a button that says something unfashionable and delivers exactly that.
If you are building a downloader, check what your provider actually returns for the audio-only formats. If it is M4A or Opus, label it that way. Your users' devices already play it, and you get to skip the entire transcoding headache on day one.