INTERACT FORUM

More => Old Versions => JRiver Media Center 21 for Windows => Topic started by: kstuart on November 19, 2015, 06:18:57 pm

Title: Conversion of 96khz audio files to 44.1khz ?
Post by: kstuart on November 19, 2015, 06:18:57 pm

So, does MediaCenter have a way to convert 96khz files to 44.1khz without approximations?

Clearly, if one was going from 96khz to 48khz, you just throw out every other sample.

But, to go to 44.1khz, there seems to be a lot of math involved.  Conceptually, it seems that it involves some points where - for example - the calculated sample is 12.37 and so you have to round down to 12.

Or, am I missing some advanced math tricks that are involved ?

Thanks.

Title: Re: Conversion of 96khz audio files to 44.1khz ?
Post by: blgentry on November 19, 2015, 06:24:14 pm

If my minimal understanding is correct, going from 96k to 48k does NOT just throw away every other sample.  Dither is applied (noise is added) because somehow just throwing away samples makes things sound bad.  I think this is referred to as aliasing.  But my digital audio theory needs work.

So, as I understand it, the "big math" is done ANY time there's a sample rate conversion in MC.  Hendrik schooled me on this a little while ago.

Brian.

PS:  Is that you over in the head-fi forums with a John Coltrane album cover as your avatar?  A Love Supreme is one of my favorite albums of all time, even though I'm not a very big Jazz fan.

Title: Re: Conversion of 96khz audio files to 44.1khz ?
Post by: ferday on November 19, 2015, 09:46:45 pm

http://www.dsprelated.com/freebooks/mdft/Sampling_Theory.html

Some basic sampling rate math.  Thank goodness computers are pretty good at summing things

Title: Re:
Post by: Trumpetguy on November 20, 2015, 05:46:29 am

No, you do not discard every n'th sample when downsampling when the sample ratios are whole integers. At least I hope MC resampling is more sophisticated than that. You do the proper filtering an discarding or adding samples in any resample process.

Title: Re: Conversion of 96khz audio files to 44.1khz ?
Post by: dtc on November 20, 2015, 07:04:21 am

Upsampling and downsampling is a process of interpolating plus filtering. The filtering is necessary to remove anomalies produced by the interpolation. Even if you just take every other sample (96KHz to 48 KHz) there will be aliasing noise introduced which needs to be filtered out. None of the math of this is easy, but here is a relatively approachable simplified explanation of the basic interpolation and filtering process - approachable, that is, if you are pretty good at some advanced mathematics.

 Multirate DSP, part 1: Upsampling and downsampling (http://www.eetimes.com/document.asp?doc_id=1275556)

Multirate DSP, part 2: Noninteger sampling factors (http://www.eetimes.com/document.asp?doc_id=1275559)