Best 24 bit Jtest .wav files?

[Pacific.Digital]

Does anyone have a 24 bit JTEST .wav they'd be willing to share? Some are available online but their provenance is a bit murky.

 

 

[tpaxadpom]

If you have access to AP 2722 you should be able to create one. I generated one for 16bit/44.1kHz but never had a need for 24bit version. You should specify sample rate you want. If I get a chance I will create one.

 

[1audio]

I could use one as well. I have ones I generated but I don't trust the provenance (me), and would at least like to have the same as the rest of the industry or at least a few others as alternate references.

 

I find I need a long file (min 5 minutes) since the long FFT (16M) I use takes a lot of samples. I have used Audacity to extend the ones I generated but I'm not sure that they are still "perfect". I do get good results.

 

[PeterSt]

Count me in as one liking to have that. But just as precautious you are on having it right, I wouldn't think of creating it my own (which actually is easy in the base).

 

Possibly changing the subject (but only a little), do we know how difficult it already is to have a decent normal testsignal in the first place ?

Never just grab something which doesn't come from a real good source. You could compare different testsignals, and take the one which produces the best test figures (on random DACs I think).

 

But I wouldn't dare to go that route for JTest ...

 

Peter

 

PS: Demian, supposed you found a good one but it is too short for you ... couldn't you use some Repeat function in the player you use ? If you think that would solve it, but don't know this player, use XXHighEnd (A-B function). BUT, first take care of the proper alignment of the file for its start and end (the end must match the start). You'd have to do this in a wave editor. But then this will be quite the same as your extension of the Audacity file ... Quite, because it might be tough to find the original end point hence find the connection (which of course is at the sample level).

 

[1audio]

I am not sure how the repeat function will splice beginning to end. Testing it is too tedious for me. Splicing with Audacity works and I think I lined up the segments perfectly but I'm not sure.

 

A little chunk of software that generated specific test files would be great (hint, hint!). There are only a few test signals that really need this treatment- the JTest for the different sample rates. Test tones that are precise sub multiples of the sample rates as well for jitter tests and impulse files that allow 1 and 4 sample impulses for example.

 

I use the long file and high resolution FFT to look at the "close in" phase noise and get a very low noise floor. It also reveals ASRC's which all seem to have some sub 10 Hz sidebands.

 

[Pacific.Digital]

@tpaxadpom... thanks so much. A 44.1kHz/24 bit would be great to start, and then a 176 and 192 thereafter. Some devices are highly data pattern sensitive, others seem to be immune. We have the APx515 which will play .WAV, but it doesn't seem to have the ability go create arbitrary data files.

 

[Pacific.Digital]

Demian,

 

Here's two for you to try out, along with the C code which should allow you to make your own files.. They're from WakiBaki at DIY Audio...

 

Jitter2 is 48/24 and Jitter3 is 44/24.

 

http://www.diyaudio.com/forums/digital-source/183110-jitter-test-signal-j-test-signal-matlab-3.html

 

Should produce 1 minute of 24 bit JTEST...

 

http://www.ne.jp/asahi/fa/efu/soft/ws/WS140.ZIP for a 44/16

 

[PeterSt]

I am not sure how the repeat function will splice beginning to end.

 

Maybe I misunderstand you, but what I wanted to say is that XXHE is explicitly made for just this (not missing or duplicating even one sample etc.). I use it myself for measuring, so ...

(so you don't need to worry about *that* part)

 

[PeterSt]

Here's two for you to try out, along with the C code which should allow you to make your own files.. They're from WakiBaki at DIY Audio...

 

Before I say what I intend to, let me first say that this is not about the competence of WakiBaki, just as I respect all people there. But :

 

This would be the typicle example of something I wouldn't like to use much. Just read through that couple of pages overthere, and you'll know what I mean.

Additionally (and I did NOT sort it out really), to me it looks like this is outputting at -0dBFS. Not good ...

 

Also Demian, I think if one is really experienced on this you may get something from this when it were about i2s. But keep in mind that JTest is about data correlated jitter. Not the most useful for i2s ...

(I only say this because I have the hunch you are into i2s :-)

 

Regards,

Peter

 

[1audio]

maybe this weekend I'll try the files and see if I get different results from my other files.

 

They are full scale which is normal to maximize the SNR for the measurements.

 

I have seen meaningful output from this test in both I2S only systems and systems with SPDIF in the chain. However it really only measures steady state conditions and doesn't really show the impact of constantly changing audio modulating clocks.

 

I have been working on ways to measure jitter down to very low levels with a goal of a way to see (and possibly hear) the real time jitter inside a system. I can see the degradation of the clocks as they get passed from stage to stage. Calibration has been the challenge moving from phase to frequency to time and looking at deterministic vs. random jitter. Its been a very interesting learning experience.