Home Propeller Head Plaza

Technical and scientific discussion of amps, cables and other topics.

Re: Why, a lot of good questions...

OK,

I looked at the article in question and it brings up a few questions:

1) There is nothing stating what the quantization level is - since the y axis is labelled -1 to 1, it suggests to me the levels are floating point (were they converted to this after the integer quantization is done or was integer quantization even done? I imagine it is but it not stated anywhere in the post that it was 16 bit)

2) 1 degree of phase is still a large phase shift - do we even know what is discernable by the ear? I imagine it is frequency dependent too.

3) The linear interpolation between points doesn't tell the whole story and massages the differences visually. I would love to see (a) all the 8 phases from 0 to 7 on the same axis, (b) the original analog signal sampled and the output after the reconstruction filter for each phase on the same axis, and (c) the 8 phases after reconstruction filter on the same axis. If you change the values of two points slightly the line between them looks almost the same, if you put it through a reconstruction filter the differences should become clearer.

I am not trying to discount any of this - I am just being overly skeptical until I get all the details of these results. If that is 16 bit sampling it does seem that there is pretty darn good phase accuracy even at 20Khz.


This post is made possible by the generous support of people like you and our sponsors:
  Sonic Craft  


Follow Ups Full Thread
Follow Ups


You can not post to an archived thread.