Clarinet Notes

Odds and ends of clarinet tech and training

Does YouTube reduce dynamic range of recordings?


This is part of an exchange I had with some folks on Clarinet Transformation Community

 I did some investigation by googling, plus examining one of my own videos.  

  • Yes, YouTube does modify videos. There’s reference to “compression”. Unfortunately, that word has two different meanings. Compressing to make the file smaller by reducing video resolution has no effect on audio. The other kind of compression adjusts the audio by equalizing highs and lows. That we don’t want. My impression from text I read was that YT was definitely compressing video. They are probably adjusting audio loudness. That may be a step adjustment that has nothing to do with range. They may be compressing audio.
  • YouTube also presents different content based on the platform. In other words, on a phone they may be saving bandwidth by returning lower quality content. Or the device receiving the content may be asking for low quality. So what I see/hear on my wired-internet desktop computer may be different from what you would hear on a WiFi iPad or a 5G iPhone.
  • I did an experiment with one of my own files. I did some stats on 1) the original, 2) the thing I got back when I went to YT studio and asked to download my upload, and 3) the thing I got  when I asked a streaming app to download my video. Here’s the result

Wow, so if I’m translating your findings into Tish-talk does that mean: yes, the youtube playback (stream?) compresses the dynamic range by one-half its original?? 

 Yes, in this case, 50% is what the numbers said to me too. Not sure the file I chose was the best example. Vimeo purports not to change content. However, the software that plays it on your computer might have some influence too. Not sure I can measure Dynamic Range on Vimeo output though. My tool only works with YouTube. I can DVR the recording and measure that, but I’m introducing another source of error when I do that.

I’m going to get philosophical for a sec. Speculating a bit, I think the three of us have different sound goals, so we listen for different things. 

  • Tish mostly plays live. She wants her recordings to sound realistic — exactly like she sounds live so she can get better feedback. In theory, that means a mic that captures reality as much as possible (including gain set to best record the dynamic range of her performance), a dead room with no reverb, and no digital enhancements (like what’s done in Dolby On).
  • I never play live, and I want my recordings to sound better than the real me. Particularly better sound quality. The first thing I listen for is tone. Dynamic range is secondary. For my recording setup I chose the best quality studio mic that I could afford, aiming for a dead recording space and planning to do digital enhancements such as reverb and EQ. I was not recording video at all, but I wanted maximum lossless audio with maximum bitrate. And Mic gain set as I advised Tish. I was mixing with accompaniment tracks. BTW, I did this setup long before I was doing recordings for Transformation Community. When I started recording video, I think I lost some audio quality. 
  • Hiedi seems to be sort of in between Tish and me. Both of you are more concerned with dynamic range, while I am more going  for tone.

The analog sound that comes out of our clarinet goes through several digital processing steps before it comes back to our ears from a recording posted on CT. Any of these steps can change quality (bitrate, digital range, and other things as well)

  1. Analog sound from our clarinet (mixed with room/background noise) goes to
  2. Mic, which sends digital signal to
  3. Computer’s sound card, sends digital signal to
  4. Software that creates digital file MP4, MP3, etc. Unless you produce WAV audio, this file is compressed for space. The codec of the output format has some effect here.  Any VST effects you apply in software also affect quality.  These effects may make it “better”, but they always make it different.
  5. You upload the file to some service that stores it. Uploading won’t change it. However, the service may compress it to save space on servers or to standardize quality.
  6. Software on the listener’s computer requests the file to play it back. Either the requesting software or the server may modify the file to save resources.
  7. Speakers or headphones on the listener’s computer play the sound. 

That’s a lot of places where quality — including dynamic range — can get lost along the way.

(Sorry to be so long-winded)

Leave a Reply