Dowsampling question (integer vs non-integer ratio)

[Jakub]

Hello. 

I am looking for some tips about downsampling material recorded in 48 or 96hz sample rate to target CD 44.1hz. My interface does not work with 88.2hz sampling rate (has only fixed 44.1, 48, 96 and 192). I read some mixed information about integer vs non integer ratio with downsampling. As I do not have a possibility to record in 88.2 integer downsampling to 44.1 is not possible. However I also read that modern downsampling use complex algoritms so the integer vs non-integer thing is no longer relevant (one quote I found : 

Quote

Sampling rate is not simply dropping every other sample, etc. It's actuality done by taking the master clock rate (sampling rate like 88.2kS/s) and upsampling the data to over 30 MEGA samples per second. Once it is has been upsampled to that super high sampling rate, then it's downsampled to the appropriate sampling rate.

So - can we generally say that in Cakewalk by Bandlab it is safe to downsample from 48 or 96 to 44.1 and we should not have any problems/distortion etc. due to ratio of two sampling rates being non-integer?

[John Vere]

Interesting question and one I’ve not heard before. Lots of discussion about bit depth and dithering but I’ve never given how clock rate might be converted any thoughts. 
 

I record at 48/24 and export all my songs as such. I then batch convert to MP3 320 using Gold Wave Editor. 

 For burning a CD I just drag the 48/24 album into Nero and it converts for me.

But now you’re got me thinking of this:

Does all software use the same quality algorithm to convert clock rate?   Not that the CD’s I burn will be noticeably different because these are just for proofing mixes. 

But this would be important information for a proper production run of real CD’s so a good question to ask the replication people. 

I think the answer will be to please submit a 44.1/16 master. Which brings us back to your question!
Does Cakewalks Export function do the job properly or would Wave Lab or ?   be better?   

Edited May 26, 2022 by John Vere

[bvideo]

Quote

Sampling rate is not simply dropping every other sample, etc. It's actuality done by taking the master clock rate (sampling rate like 88.2kS/s) and upsampling the data to over 30 MEGA samples per second. Once it is has been upsampled to that super high sampling rate, then it's downsampled to the appropriate sampling rate.

That quote fails to mention the quality of the algorithm used for upsampling or any bandpass filtering (also assuming they mean downsampling is taking every n'th sample). Better to quick read over the wikipedia article for an idea of digital resampling. Also, there's this site http://src.infinitewave.ca/ that offers comparisons of various daw results for 96KHz to 44.1KHz resampling.

Bottom line from infinitewave* is Sonar X3 downsampling, while not being the very very best of all possible digital SRC algorithms for all test scenarios, doesn't apparently have audible artifacts. (They don't seem to have CbB explicitly listed.)

(* Magix Vegas 17 yikes!)

Edited May 26, 2022 by bvideo
new info

[Quick Math]

For 3rd party Tools that are probably better than DAW in terms of quality, Voxengo offers free SRC software called r8brain.

https://www.voxengo.com/product/r8brain/

Other than that, RX (iZotope) and Saracon are often mentioned for the best SRC out there but these are not free.

 

 

 

[bitflipper]

r8brain is great for batch conversions, e.g. you want to convert a whole folder full of files at one go. It's also a very good converter, perhaps the best out there. There is a free version, which is open-source if you want to totally geek out.

Cakewalk's conversion is also one of the best out there. Although it's not included in the SRC comparisons that bvideo linked to above, I don't think the algorithm has changed since SONAR. I have r8brain and several other applications that all do great sample rate conversions, but I never bother going outside Cakewalk to do it unless I've got a whole bunch of files to convert.

Bear in mind when comparing SRC converters that being able to see artifacts in the graph does not necessarily mean you can hear them. You will not hear artifacts from Cakewalk's SRC. The difference between CW's and r8brain's aliasing is about 5 dB, with both down around -170dB, well below the threshold of audibility.

Bottom line: let Cakewalk do the conversion, it'll be fine. 

[John Vere]

Interesting info. I sort of thought this myself and have never doubted that Cakewalk probably used good convertion. 

Also Interesting that the Op left the building even though it shows they logged on 10 hours ago? 

Edited May 31, 2022 by John Vere