Please familiarise yourself with the forum, including policy on feature requests, rules & guidelines

Sample interpolation types - what's the real difference?

1
rudolphrapidrudolphrapid Beta Tester Posts: 129

I decided to open a separate thread for this. I'm really wondering what the difference between the two interpolation option LINEar and SINC is. I compared them to each other using many different samples from drones to drum loops to complex tracks and I barely noticed any difference when stretching and/or pitch shifting. I set pitch and speed INDEpendent, of course. If one produced artifacts the other also produced almost the same and if one sounded good so the other did.

The only big difference I managed to notice is that by SINC I got sound drop many times :) (So the sample stopped playing back in the middle or even at the very beginning of the sample because of bandwidth/processing power overload, I think, which was strange.)

What are the expectations with SINC over LINEar?

Comments

  • 0
    rudolphrapidrudolphrapid Beta Tester Posts: 129

    Small clarification: with SINC I usually get sound drops when the unison is turned on with number of 3 or more.

  • 0
    amiga909amiga909 Central EuropePosts: 1,078

    @rudolphrapid said:
    Small clarification: with SINC I usually get sound drops when the unison is turned on with number of 3 or more.

    quickly tried, how can you reproduce that? that would be a bug.
    about SINC/LINEar: SINC is newer and better, i can clearly hear a difference when a sample gets pitched/stretched (you might need to try extreme settings to hear it). afaik LINEar is really just there because of backwards compatibility.

  • 0
    rudolphrapidrudolphrapid Beta Tester Posts: 129
    edited August 2019

    Here's an example. I sampled this internally so the file format must be the best for Deluge. Just strecthed the sample (mode set to stretched and drew a 4 bar long note in the clip).

    LINEar: https://www.dropbox.com/s/kj1ip2v5bn7jrbv/stretched_linear.wav?dl=0
    SINC: https://www.dropbox.com/s/ggqs3crv1euyd2i/stretched_sinc.wav?dl=0

    Do you hear difference? :) It's really slight if any.

    The difference may be a bit more recognizable when pitching the sound one octave higher.
    LINEar: https://www.dropbox.com/s/de4gng253qcvwpe/stretched_pitched-up_linear.wav?dl=0
    SINC: https://www.dropbox.com/s/4ld7vgspmwprknv/stretched_pitched-up_sinc.wav?dl=0

    And even with a single kit having this single stretched sample in it I managed to create sound drop off by increasing the parameter number on unison.

    Post edited by rudolphrapid on
  • 0
    amiga909amiga909 Central EuropePosts: 1,078

    @rudolphrapid said:

    And even with a single kit having this single stretched sample in it I managed to create sound drop off by increasing the parameter number on unison.

    Thanks rudolph, yeah, I agree the difference is subtle. Some more explanation in the manual would be nice.
    I'll try the reproduce it later. Whats the sample you are loading (length of sample, resolution, size in MB)? I am certain Rohan will be very pleased to get the XML (you can send him a mail with the XML and best with sampled attached via collect media), if you can get a reproducible case for sound drop outs with samples. I have experienced this many times before but never could nail it down in a simple song.

  • 0
    rudolphrapidrudolphrapid Beta Tester Posts: 129
    edited August 2019

    I created a test song you can check if you have time: https://www.dropbox.com/s/prxonvhywgz5a2y/timestretching_test_song.zip?dl=0

    (I can't find interpolation type in the xml...)

    Thanks for your help.

    Post edited by rudolphrapid on
  • 0
    amiga909amiga909 Central EuropePosts: 1,078

    thanks @rudolphrapid, will try in the evening. if this is reproducible, you should get a price for the bug finding of the month! or even of the year! :) btw, strong bug finding skills are much appreciated in the v3 beta facebook group (yeah FB sadly but better than nothing).

  • 0
    amiga909amiga909 Central EuropePosts: 1,078

    checked in current v2 release version and newest v3 beta. nice sample :)
    how do you get an audio dropout? when i hit Play it plays fine for at least 1min.

  • 0
    rudolphrapidrudolphrapid Beta Tester Posts: 129
    edited August 2019

    It's not predictable when it starts drop outs. You can play with unison parameters heavily a bit and change pitch in parallel.
    Last time I got drop outs just because rising the pitch and lowering the tempo a lot.

    Anyway do you hear differences between the interpolation types?

    The sample is from Sting, so yeah, it's nice :)

    Post edited by rudolphrapid on
  • 0
    amiga909amiga909 Central EuropePosts: 1,078

    unison is very cpu heavy and may not be suited for changing the value very fast.
    yeah, i cant hear a difference either. dont think it has musical use, just leave it at SINC. when the option was introduced it was meant to replace Linear and Linear was kept for backwards compatibility. dont have a clue in which situations it makes what kind of difference, but i guess Rohan had reasons to add the option.

  • 0
    rudolphrapidrudolphrapid Beta Tester Posts: 129
    edited August 2019

    I usually don't change unison parameters crazy. However it took me some to hunt down what happened when I got strange sound drop outs even with 2-3 "basic" tracks going on :)

    There's so minimal info available about new time stretching algorithm except the single advertisement in 2.1 patch notes that really takes expectations high :)

    Post edited by rudolphrapid on
Sign In or Register to comment.