Forums

DSP; Framing and Windowing in Audio Signal Processing

Started by sparrowJr 5 years ago6 replieslatest reply 5 years ago1121 views

Post deleted by author

[ - ]
Reply by Rick LyonsJuly 7, 2018

Hello sparrowjr.

Someone here may be able to help you if you define (as clearly and completely as you are able) EXACTLY what your words: "make a new signal", "delay-line based Method", "frame", "framing", "window", "windowing", "better" (used twice), "Overlap", and "framestep" mean.


[ - ]
Reply by sparrowJrJuly 11, 2018

Hello, Sir..

sorry..

*make a new signal: synthesis signal from my audio data (pitch shift my audio) by delay-line method.

*delay-line method: one the methods to make a new signal SOLA, Psola and Phase Vocoder.

so..

I am trying to make a synthesis signal from my audio data (pitch shift my audio) by delay-line method. As I know, the implementation of delay-line method  is going on the time domain processing. My question is, should I framing and windowing My audio data before trying to pitch shifting it?

Thank You, Sir 

[ - ]
Reply by wnoliveiraJuly 12, 2018
Hi sparrowJr,

As said before, windows are commonly used when you process data in frequency domain.

Delay lines can be implemented with simple buffering in time domain. So, you need to check your impplementation method to see if it uses some processing on frequency domain or just in time domain. The formulation of method should give you a good idea of how to implement it.
[ - ]
Reply by dszaboJuly 7, 2018

Could you elaborate on what you are trying to do?  When you say “delay line based method” it sounds like you are making a filter.  Filters do not use windowing or frames, unless they are being implemented in the frequency domain, which you’ve said explicitly you are not.  I’m confused because I would not equate filtering and synthesis

[ - ]
Reply by sparrowJrJuly 11, 2018

Thank You for Your attention.

I am trying to make a synthesis signal from my audio data (pitch shift my audio) by delay-line method. As I know, the implementation of delay-line method  is going on the time domain processing. My question is, should I framing and windowing My audio data before trying to pitch shifting it?

[ - ]
Reply by dszaboJuly 11, 2018

Here’s an article detailing an implementation: http://msp.ucsd.edu/techniques/v0.11/book-html/node115.html

It was the first hit in a google search.  I skimmed it and it seems to be consistent with how I’ve seen this done before.  Still not sure what you’re asking so I’ll stick with my original “no”.  I just don’t see what framing or windowing has to do with it at all.  That being said, there is a window associated with the delay line to eliminate discontinuities when the tap rolls from the end to the beginning.  I don’t think there’s a correct window to use, but my gut would say triangle.