SRC and Dithering settings (in general)*

Steve, never mind, I just found how to downsample using the Foobar GUI.

It sure beats using the command line.

 

Grant,

Here's a sample from a Nick Drake tune of a laserdrop I did. I downsampled it directly from the original 192/24 file to 44.1/24 with SoX using Foobar. And, to follow your same process, I then applied MBIT+ dithering to it tp take it to 16 bits. The only 'bad' thing I did to the file was add fade in and fade out via Audition when it was already at 44.1/16 file: http://www.mediafire.com/?0nl0zjmgwky

Do you hear any artifacts on it?

 

Your sample serves no purpose if I can't hear the original.

 

Hi Grant,

Are you considering the limitations of 16/44 vs. a high res original?
Leaving aside issues of artifacts from the processes, the two formats isn't tiny. (I'm sure you know this.)

Also consider the filtering in the digital converter. The filter used at a high sampling rate is gentler and in many ways, places less demand on the designer.

Then again, you've said each process by itself sounds fine.
And using RX as a standalone leaves aside any issue of that final save in Foobar.

Sounds like a mystery to me.

Best regards,
Barry
www.soundkeeperrecordings.com
www.barrydiamentaudio.com

 

If you want I can upload a snippet of the original, my friend.

 

Sure!

 

Grant, here's a snippet of the original 192/24 file: http://www.mediafire.com/?sharekey=15db0ee471f475b561d4646c62b381cbe04e75f6e8ebb871

Just one thing, since I cut both at different times, the 192/24 file starts some 3 seconds before the 44.1/16 one, although they basically cover the same part of the tune.

Let me know what you hear and/or don't hear here.

I'd also love to hear Barry take on these two files.

 

There is no file.

 

Grant, let's see if it works this time: http://www.mediafire.com/?nzum0tmeoim

 

Hi Metoo,

The audio channels are full for the next several days - a new project that needs to be finished soon.

If I can, I'll try them some time next week.

Best regards,
Barry
www.soundkeeperrecordings.com
www.barrydiamentaudio.com

 

Thanks, Barry.

I look forward to your take on the sound of the SoX SRC.

Let me just suggest that, if at all possible, you download the files now for later listen. The reason is that these online services tend to erase the files after some time.

 

Why does the original 192k sample have all that white noise on it?

 

What white noise are you referring to? Is it something you hear, or something you see in the frequency graph display in Audition? If it's the latter, what you are seeing up on top is the infamous high frequency noise that accompanies all DSD recordings. It usually starts to be seen in earnest at around 40+ kHz.

Remember, my player does not convert DSD to PCM and cuts it at around 45 kHz in playback. What you see is the whole enchilada. That is why I mentioned before that, when I am preparing 96/24 files for myself I use a no-aliasing setting to go from 192/24 to it. In this way I am sure that there is the least amount of crap getting into the downsampled file. Of course, if I went to 88.2/24 I am sure all of the basic high frequency noise would be eliminated. In fact, that is the case in most occasions even at 96/24.

 

Grant, have you done any comparisons between the two files? If so, what have you noticed?

 

BTW, now that we are talking about the high frequency noise present in DSD files. Anybody reading this and interested in knowing what it looks like if you have access to a frequency graph I suggest you download the 192/24 snippet posted before, as I realize that many do not have players that deliver DSD without converting it to PCM and downsampling to 88.2/24.

As for the owners of the classic Pioneer 563 (565 in Europe) universal player, here you have proof that it does deliver the full frequency range of DSD, noise and all.

 

Hi Metoo,

I grabbed the files, only to find they are FLACs.
If I'm going to listen to SRC and dither algorithms, I'd do this only with uncompressed AIFs or WAVs.
I'd also prefer to do it with a musical program I am very familiar with.

Listening to something I don't know, while the software simultaneously has to deal with decompressing a file is not how I would evaluate sonics.

Just my perspective.

Best regards,
Barry
www.soundkeeperrecordings.com
www.barrydiamentaudio.com

 

Hi Barry,

Thanks for downloading the files. The truth is that I uploaded them as FLAC so as not to make the up and the download to heavy. (We're usually talking a 40 to 45% smaller file than a WAV one here).

Don't you have a FLAC to WAV converter for Mac? Remember, FLAC being a lossles format if you convert the file to WAV you'll get the same quality file I have on my hard disk.

BTW, FLAC files can be decompressed to WAV without it having to be during playback. Although my computer is not a Mac I can find out what software you would need to convert FLAC file to WAV. IMHO, a FLAC to WAV to FLAC converter is not a bad tool to have at hand.

 

Hi Metoo,

I have software that will convert to WAV or AIF or anything I want.
Perhaps I'll give it a try.

Still, I wouldn't draw any conclusions regarding SRC and dither
1. with files that have undergone any additional processing whatsoever (including conversion to and from FLAC)
2. with program material I don't know very well

I could draw conclusions with material I'm very familiar with and files that have undergone only the processes being evaluated (even if they're shorter samples).

Best regards,
Barry
www.soundkeeperrecordings.com
www.barrydiamentaudio.com

 

I can understand your precautions, but FLAC is lossless and I, and others, have done MD5 testing on the resulting WAV files after converting back from FLAC and they bit exact to the originals.

If you finally decide to compare the files, even if you feel you are slightly out of your comfort zone, I would greatly appreciate to know what similarities or differences your ears hear in both files.

One thing is clear, I downsampled the 44.1/16 file in one step to 44.1/24directly from the original 192/24 WAV file. Then I applied the MBIT+ dithering to take it down to 16 bits. Finally, I converted both files to FLAC.

 

I'm not only seeing it on the graph, i'm hearing it. I guess, somehow, my soundcard settings are different than yours. I can hear more frequencies.

I can't with all that noise mixed in with the sound of the hi-rez sample.

 

Since you are using Audition, you can provide Barry with .aiff files.

 

I am totally surprised to hear this. Could it be tape hiss, or maybe something wrong with your configuration? (After all, you do mention hearing some distortion in your final downsampled and dithered files.) BTW, does your sound card do 192/24? Have you tested it at this sample rate before?

Now I am more interested thatn before in knowing what Barry hears on - at least - the 192/24 file, because I am playing it now and do not hear any white noise on it. Could it be something to do with the way your set up handles 192/24 FLAC files?

Please verify the 192/24 FLAC file you downloaded with this file verify to see if what you have is exactly like the file I have on my hard disk: http://www.mediafire.com/?sharekey=2c2cefdfbf69a08cd0d290dca69ceb5ce04e75f6e8ebb871 If the checksum does not find them the same I will re-upload it to a different service.

BTW, has anyone else here downloaded the file? If so, do you hear the white noise Grant is talking about?

 

I will find a way to give you a sample of what I am hearing. Hang on...

 

Here ya go! I just copied the sounf from Audition to another sound recorder:

http://www.yousendit.com/download/U0d4SlIzcVg1bmgzZUE9PQ