KVR Audio

juha_p · Post by **juha_p** » Sat Aug 04, 2018 1:09 pm

Maybe developement related question ... maybe not (move if needed)?

http://www.vintageip.com/pics/markknopf ... dxmoke.jpg

On another audio related forum someone stated this is spectrum for 24-bit/96kHz audio (Foobar2000 + SPAN VST) ... if fmt is OK then, where data seen above fs/2 could come from?

Aleksey Vaneev · Post by **Aleksey Vaneev** » Sat Aug 04, 2018 1:27 pm

I think this "data" above 20 kHz may come from bad ADC electronics design.

juha_p · Post by **juha_p** » Sat Aug 04, 2018 1:33 pm

But, ... software path is Foobar2k (audio file) --> VST wrapper --> SPAN --> output and source is a 24/96 file so AFAIK, there shouldn't be data above fs/2 to play in Foobar2k ... but seems to be by SPAN. Could there be SRC somewhere (SPAN display is for 192kHz ...)?

Aleksey Vaneev · Post by **Aleksey Vaneev** » Sat Aug 04, 2018 1:37 pm

juha_p wrote:But, ... software path is Foobar2k (audio file) --> VST wrapper --> SPAN --> output and source is a 24/96 file so AFAIK, there shouldn't be data above fs/2 to play in Foobar2k ... but seems to be by SPAN. Could there be SRC somewhere (SPAN display is for 192kHz ...)?

Yes, may be a glitch introduced by SRC. But SPAN does not perform SRC, of course, it only displays what it's fed with. Sample rate is definitely equal or above 192 kHz.

juha_p · Post by **juha_p** » Sat Aug 04, 2018 1:40 pm

Does it communicate with wrapper/host somehow and ask for SRC when Mode set to <> source audio fs?

Aleksey Vaneev · Post by **Aleksey Vaneev** » Sat Aug 04, 2018 1:42 pm

juha_p wrote:Does it communicate with wrapper somehow and ask for SRC when Mode set to <> fs?

No, usual plugins can't "communicate" sample rate, plugins obey sample rate set by the host.

juha_p · Post by **juha_p** » Sat Aug 04, 2018 1:44 pm

So, when you set SPAN Mode to <> source fs then it just takes as much data to plot as what Mode asks?

Aleksey Vaneev · Post by **Aleksey Vaneev** » Sat Aug 04, 2018 1:58 pm

juha_p wrote:So, when you set SPAN Mode to <> source fs then it just takes as much data to plot as what Mode asks?

"96kHz" preset in SPAN simply changes display range, it does not affect sample rate.

juha_p · Post by **juha_p** » Sat Aug 04, 2018 2:13 pm

Yes, ... but ... when the display range is set to 96kHz (which I understand is best for 192kHz source) and the source file is 96kHz as well then where SPAN finds data for up to source fs ... ?

Aleksey Vaneev · Post by **Aleksey Vaneev** » Sat Aug 04, 2018 2:14 pm

juha_p wrote:Yes, ... but ... when the display range is set to 96kHz (which I understand is best for 192kHz source) and the source file is 96kHz as well then where SPAN finds data for up to source fs ... ?

It does not find it, if the source file was really 96kHz SPAN would display a red line at 48 kHz frequency.

juha_p · Post by **juha_p** » Sat Aug 04, 2018 2:16 pm

Aleksey Vaneev wrote:
juha_p wrote:Yes, ... but ... when the display range is set to 96kHz (which I understand is best for 192kHz source) and the source file is 96kHz as well then where SPAN finds data for up to source fs ... ?
It does not find it, if the source file was really 96kHz SPAN would display a red line at 48 kHz frequency.

That is what I was expecting but somehow it does not ... (my raw guess was that garbage there is left in buffer initialization (if not "zeroet") or some other "memory related" issue?).

I have to look the VST wrapper part next.

Thanks !

BertKoor · Post by **BertKoor** » Sat Aug 04, 2018 7:26 pm

juha_p wrote:

On another audio related forum someone stated this is spectrum for 24-bit/96kHz audio (Foobar2000 + SPAN VST) ... if fmt is OK then, where data seen above fs/2 could come from?

Take one step back please. What exactly are we looking at? What track is playing in foobar? What is the playback device?

If the music is a regular 44.1 kHz mp3 or flac or whatever, and the sound card is set to 96kHz, then it is probably FooBar which does arbitrary SRC and you're seeing its artifacts.

But the music is like a SACD already, then you're looking at the easter eggs the mastering engineer has put in for bats & dogs.

juha_p · Post by **juha_p** » Sat Aug 04, 2018 8:28 pm

By information posted with that plot image on another forum, source file was 24-bit/96kHz file bought from HDTracks. There were few other 24/96 files plotted by same guy ... all had the same 'issue'.

As pointed already a 96kHz audio file can hold up to 48kHz (fs/2) so all data above this limitation one can see in plot are not from the source audio file but are from some other source ... either from foobar2k, VST Wrapper or SPAN ... (or because of the source audio file format has been 24/192 instead of claimed 24/96).

Now it looks clear that SPAN does not change the data for spectrum plot so it has to be from other source.
IIRC, SRC in foobar2k is done by SRC plug-in (good enough to not produce artificas like that http://src.infinitewave.ca/ ).

JCJR · Post by **JCJR** » Sun Aug 05, 2018 3:40 am

Not specifically relevant to the topic, but SPAN is one of my favorite plugins. Very useful well-done tool.

juha_p · Post by **juha_p** » Sun Aug 05, 2018 8:26 am

JCJR wrote:Not specifically relevant to the topic, but SPAN is one of my favorite plugins. Very useful well-done tool.

Who could disagree that.

Well, as I'm on Ubuntu so, I had to install Windows 10 Preview into VirtuaBox to get tested something ... .

Found out that the guy behind the graph I posted here must have used some low quality SRC plug-in before SPAN. I tried some older version of Secret Rabbit based plug-in with Linear and ZOH interpolation methods and both of those resulted simliarly and produced garbage at high frequency area ... just like in plot image I shared (excluding those sharp spikes). Other selectable interpolation methods (all Sinc based) did not screw the data.

So, I think this is solved now.

Voxengo SPAN, spectrum question