Using the search means you accept our Terms of Use.

Search
YouTube conversion

How YouTube Compresses Audio, From Upload to Playback

YouTube never plays the file a creator uploaded. It plays compressed AAC or Opus copies, and those copies are the real source of any MP3 made from a video.

5 min read · Published · Updated

Published by MP3Juice under our editorial policy.

Headline "Inside YouTube audio" beside a video player whose waveform passes through filters into two separate audio streams
Contents
  1. Does YouTube compress audio? Every upload is encoded again
  2. The streams YouTube makes
  3. What YouTube audio compression removes
  4. Sound and picture are stored separately
  5. Loudness is changed in the player, not in the stream
  6. What this means for an MP3 made from a video

YouTube re-encodes the soundtrack of every upload into AAC and Opus streams that run at roughly 128 to 160 kbps for most videos, and those streams are what you hear and what any converter receives. By then the audio has usually been compressed twice: once when the creator exported the video and again after upload. Turning it into an MP3 compresses it a third time.

Knowing how YouTube audio is compressed clears up most surprises with converted files: a 320 kbps MP3 that sounds no better than the video, a 4K upload that sounds thin, a download that plays louder than the stream. Those are the files a YouTube to MP3 converter hands you.

Does YouTube compress audio? Every upload is encoded again

Yes. A creator uploads one file. YouTube's published advice asks for AAC-LC or Opus audio at a sample rate of 48 kHz, with a bitrate of 384 kbps for stereo. That is far more data than any listener receives. The generous figure is deliberate: the cleaner the input, the less audible the damage after YouTube's own encode.

After upload, YouTube decodes the soundtrack and produces several new versions for different devices and connection speeds. This happens whatever the creator sends: a lossless soundtrack and a heavily compressed one come out as the same set of streams, though not with the same quality. The uploaded file itself is never streamed. This is also why a brand-new video may play only in low quality for its first minutes. The better versions are still being made.

The streams YouTube makes

Two codecs do nearly all the work. AAC is stored in an MP4 container and plays on practically every device. Opus is stored in WebM and is the newer and more efficient of the two. For most videos the main streams run at roughly 128 to 160 kbps, with smaller versions kept for slow connections. Which one you hear depends on the browser or app, a point covered in YouTube Opus vs AAC. The YouTube Music app is a separate case: Premium members can choose an audio quality setting with an upper bound of 256 kbps.

The two versions also differ in sample rate. The AAC stream is usually 44.1 kHz and the Opus stream 48 kHz. Either is enough to hold every frequency the ear can detect, so the higher figure is not a mark of better sound, and a converter gains nothing by raising it.

The table puts the data rates of four stages side by side:

StageTypical formatData rate for stereo
Studio masterUncompressed PCM1,411 kbps at CD quality
Creator's uploadAAC-LC or Opus384 kbps recommended by YouTube
YouTube streamAAC or OpusRoughly 128 to 160 kbps
Converted MP3MP3128 to 320 kbps, as chosen

The stream carries about a tenth of the data on a CD. The last row is the odd one out: it is the only stage where the number can go up again, and a bigger number there does not bring back anything the row above removed.

What YouTube audio compression removes

AAC and Opus are both lossy. They cut the signal into slices of roughly 20 milliseconds and work out which parts of each slice the ear will notice. A quiet sound next to a loud one at a similar pitch is masked, so the encoder stores it coarsely or drops it. The very highest frequencies are often trimmed. Sound that is identical in the left and right channels is stored once instead of twice. How AAC compares with MP3 is set out in MP3 vs AAC.

How much is removed depends on the bitrate. At 128 kbps the encoder has 16 kilobytes for each second of stereo sound, where a CD spends about 176. Everything it keeps has to fit inside that budget.

At these bitrates the result is close to the original for most music. The weak spots are predictable: dense cymbals, applause, harpsichord, anything with a lot of fast, bright detail. A poor upload makes things worse. If the creator's file was already a low-bitrate copy, YouTube's encoder spends its data faithfully preserving the first encoder's mistakes.

Sound and picture are stored separately

YouTube uses adaptive streaming. The player downloads picture and sound as two separate streams and keeps them in step, switching the picture between resolutions as your connection changes. The audio stream is chosen on its own terms. A 4K label therefore says nothing about how the soundtrack was recorded or encoded, and a lyric video with a still image can sound every bit as good as a cinematic one.

The same separation keeps audio conversion quick. A service that needs only the sound can fetch the audio stream, a few megabytes for a typical song, and ignore the far larger video stream.

Loudness is changed in the player, not in the stream

YouTube also evens out volume between videos. When a track is mastered louder than the platform's target, the player turns it down, and Stats for nerds reports the amount as content loudness. That adjustment happens during playback. The stored stream keeps its original level, so a file converted from it can play noticeably louder than the video did on YouTube. Nothing went wrong in the conversion. Other causes are listed on our page about why a downloaded MP3 sounds different.

What this means for an MP3 made from a video

Converting to MP3 adds one more lossy generation, because the conversion service has to decode the AAC or Opus stream and encode the result as MP3. That work is followed in the five stages of a video to MP3 conversion. In practice, five points follow:

  • The stream is the ceiling. Asking for 320 kbps from a stream of 128 kbps gives a bigger file with the same sound.
  • A moderate setting such as 192 kbps is normally enough to keep the extra loss inaudible.
  • The upload matters more than any setting. Prefer the artist's official channel to re-uploads, sped-up edits and recordings of a screen.
  • Ignore picture resolution when judging sound.
  • Listen before you save. On MP3Juice, the Play button on each result opens the video so you can check it.

None of this makes a converted file a poor one. A clean official upload, converted once at a sensible bitrate, sounds very close to the stream on ordinary equipment. The wider process, for MP4 as well as MP3, is in our guide to YouTube audio and video conversion.