Topic: Resampling Algorithms compared

[jerryfreak]

check it out, old article, but new to me:

http://www.bias-inc.com/products/peakPro5/resampling/peakResamplingWhitePaper.pdf

I've been using soundforge, which isnt the best (though i bet the errors are inaudible.

incidentally, wavelab looks really really bad.

I think i'll use audition from here on out.

I'd love to see a similar comparison of dither algorithms, although thats maybe a bit more subjective, where resampling is all about *accuracy*

[libfab]

Interesting article, with an embedded link to http://src.infinitewave.ca/ .
It would be interesting to test a few open source apps like SSRC, which is reputedly better than SRC, or PPHS.

[twatts (pants are so over-rated...)]

I used to use SF5.0...  Now I use Audacity, though I'm still getting used to it...

I'm curious as to the consensus here, as I've jsut entered the 24bit realm...

Terry

[Nicola Fankhauser]

hi

graphical evidence seems pretty clear, however it's not clear at all how this translates into audible problems. sometimes less precise algorithms produce more appealing results (dithering for example).

I'd love to see audacity in the comparison, though.

regards
nicola

[Roving Sign]

I think Brian did some informal comparisons a while back - should be a link around...

[JackoRoses]

check it out, old article, but new to me:

http://www.bias-inc.com/products/peakPro5/resampling/peakResamplingWhitePaper.pdf

I've been using soundforge, which isnt the best (though i bet the errors are inaudible.

incidentally, wavelab looks really really bad.

I think i'll use audition from here on out.

I'd love to see a similar comparison of dither algorithms, although thats maybe a bit more subjective, where resampling is all about *accuracy*

than you so much for this link. I say the pic on WL4 really takes the wind out of all the fluffing it gets. I use it myself because I haven't heard anyone say anything bad about it really. I would like to see a WL5 comparison but I don't think the difference would be much. The "whats new" pdf file that comes with WL5 makes no mention of reworking or improving any algorithms.
I'm going to look into adobe audition right now.

[Brian Skalinder]

I'd love to see a similar comparison of dither algorithms, although thats maybe a bit more subjective, where resampling is all about *accuracy*

A while back, I posted a dither comp I performed:  http://taperssection.com/index.php/topic,51478.0.html

[jerryfreak]

just to be clear, i expect resampling to be 'transparent', it shouldnt add a flavor.

Dither, on the other hand, can make a source sound different (better?), although its primary task should be to translate harmonics and imaging from the -96 to -115 dB range into an audible 16-bit waveform, while masking any rounding errors into the LSB of the resulting waveform. All while remaining as true as possible to the original 24-bit file.

Thanks for the links brian. I'm gonna try to do a shootout of the different algorithms i have available to me:

3 settings on the ad2k
3 settings in soundforge
wavelab
audition

I'll normalize the waveforms to -40 dB peak to try to bring out the sound of the dither


unfortunately, i cant include the v3's ANSR as theres no digi-in

[stirinthesauce]

just to clarify on wavelab.  There are 3 types of algorythms for dithering.  The internal wavelab, apogees old uv22 and their renowned uv22hr.  I'm betting (though not positive) they were testing wavelabs own (internal) algorythm.  If using apogee's uv22hr, this is a proven standard...with I'm sure better algorythms out there.  Just wanted to post this so someone doesn't assume that wavelab is poor, just don't use their algorythm.  And why when you have apogee's dithering scheme available?

[JackoRoses]

Yes but it also was testing the resampling part of wavlab as well which introduced a lot of noise from their tests and what other plugins are available to use besides what wavlab gives you for resampling?

[stirinthesauce]

Yes but it also was testing the resampling part of wavlab as well which introduced a lot of noise from their tests and what other plugins are available to use besides what wavlab gives you for resampling?

I'm an idiot    I was thinking dithering, not resampling. 

I record in 24/44.1 so resampling is not an issue.  For the other times, I guess I need to look at using samplitude or cubase, of which I have.  Or protools which I soon will have.