• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Samplerates, (over/re)sampling, artifacts, intermod distortion, aliasing...

Jose Hidalgo

Addicted to Fun and Learning
Joined
Aug 15, 2020
Messages
961
Likes
1,032
Location
France
Hi all, I've recently watched this video, and I've found it highly interesting and instructive so I wanted to share it :


I strongly suggest that you watch it if you want to be able to participate in this topic.

What the video shows isn't subjective : it's objective measurements. Of course, being a FabFilter video, it highlights a couple of their products, but that isn't the essential here. The important thing is that it shows three things :
  1. How some common audio artifacts are created, to the point where they reach levels that can be heard,
  2. How the samplerate can affect them (more than I thought actually),
  3. And how they can be eliminated.
Once you've watched the video, I'd like to know what you think, and what could be the consequences on our side (we are not music producers but music listeners).

The oversampling debate has been going on for some time, and I'm one of those who thinks that resampling is useless if you want to smooth the audio signal somehow, because you'll never be able to do better than the original signal. The first part of the video clearly demonstrates how the real analog signal that we all hear through our speakers/headphones is already smoothed even if the digital waveform that we see in audio editors doesn't look smooth at all. So increasing the samplerate is completely useless for that purpose : the video shows how an audio signal can be perfectly rendered (with no loss of information) up to the Nyquist frequency, which is half of the samplerate. Which proves once again that 44100 Hz is more than enough for our ears, since frequencies up to 22050 Hz are perfectly rendered.

However, the rest of the video seems to show scientifically how oversampling (on the music production side) works, and how it can reduce all kinds of audio artifacts that can potentially affect the music that we listen to : intermodulation distortion, aliasing... That part is very visual and nice to see. So that would raise the question of resampling on our side : not to further smooth the audio signal, but rather to eliminate a number of artifacts (provided they're present in the music that we listen to, so maybe that boils down to the production engineers and the tools that they used). I guess there are at least three possibilities :
  • Either the production engineers used the right tools in the right way and the music is perfect so we don't need any resampling
    (but can we be sure that 100% of the production engineers are perfect ? I would be sure of the opposite !),
  • Or the production engineers did a sloppy job and the music has artifacts, so we'd need resampling in such cases,
  • Or we need resampling anyway because the artifacts will be created in real time anytime we try to play the music.
I honestly don't know what to think about this. I merely consider myself as a "beginner with an opinion", who'll always listen to scientific evidence. Thus I'd like this discussion to remain objective, based on real arguments, and preferrably from people who've made the effort to watch this video and will be able to eventually quote/reference parts of it. If that's too much to ask, then you are also free to browse other topics on ASR.

Thanks for your understanding. :)
 
Top Bottom