[bwaslo]

Hello,

I'd like to announce that there are several audio research and measurement programs which can be downloaded from our web site at http://www.libinst.com:

PRAXIS- the well-known commercial loudspeaker and audio test and development tool, used by numerous loudspeaker designers and audio manufacturers.  The program can be downloaded free, and even if it is used without its Audpod interface and access key,  can be used for many operations including high-resolution Real Time Analyzer (RTA) and Thiele/Small measurements, script programming, response data display, formatting and printing, and very extensive signal postprocessing operations (filtering, analysis, rate conversion, etc.).  With the Audpod key (which we'd of course like to tempt you to purchase!), PRAXIS can perform a vast array of additional measurements and analyses including MLS, chirp, distortion isolation, and acoustical room parameter extraction to mention only a few.  There is also an add-on called "AudioPass" (AudPod and extra key required) which provides easily configured and used production testing of loudspeakers.
(Note: as of this date, April 2007, PRAXIS is not yet usable with VISTA)

DiffMaker - a completely freeware tool for audiophiles and/or researchers, developed out of interest in investigating what things make actual changes to audio signals.  DiffMaker provides a way to make high resolution recordings of audio equipment while the gear plays a musical or test track of your choice.  You record twice: once as a reference with the audio system in its normal state, and then again after a component has been changed.  Then DiffMaker processes the two recordings and provides a third recording that contains only the difference between the two recordings  - by listening to the track, you can hear whether the signal has perceptably changed.  The difference track is literally a result of subracting one recording from the other (after precise gain and time shift alignment).  DiffMaker is intended primarily for investigating very subtle changes for which no classically explainable audible differences are known to exist, such as with cables, tweaks, connector and component changes.  Changes that make expected frequency response and phase variations will result in very apparent difference track results due to the extreme sensitivity of the differencing process.  Also included in DiffMaker is a unique and very sensitive frequency response analyzer utilizing a special "keyed" pink noise signal to provide fast and repeatable measurements of equipment and room responses -- very useful for adjusting parametric equalizers on subwoofer channels.

We're hoping to get a "Liberty Instruments Circle" going here.  If a Liberty Instruments Circle interests you, please respond to this post to show support for the idea (we need posts to become eligible for our own circle).

Thanks,

Bill Waslo
eng, Liberty Instruments, Inc.

[Nick B]

Thanks for posting. The DiffMaker software was especially interesting to me.
Nick

[MJK]

I purchased Praxis (w/ the audiopod) this week after using LAUD for many years. I would love to see a "Liberty Instruments Circle".

[jcg744]

Hello Bill,

I really would appreciate to see a Liberty Instruments circle here.

Looking forward to see a future Vista release of Praxis...

Best regard from France

Jean-Claude

[Double Ugly]

Any plans for a Mac version, Bill?

-Jim

[bwaslo]

Hi Jim (it'd be rude to call you by your moniker!),

Sorry, no plans for Mac.  Both programs work with the deep guts of the windows multimedia system, it would be a complete restart to do it for Mac or Linux.  Also, I'm not familiar with the development tools for Mac.  Nothing against the Mac, but when all this was started it was easier to get info on audio programming in Windows than Mac.  I have had  a few reports of people using these with Win emulators in Mac, but never got the final word on what did and didn't work. 

But thanks for asking.

[Bob Reynolds]

Hi Bill,

Thanks for posting. I learned of your company/site from references made by Robert E. Greene. I believe that he uses your software in his review work. I've been pointing folks to your web site because of DiffMaker as often as I can. There should be a post here at AC.

I've wondered how in using DiffMaker you keep the test sequences in sync with each other? I mean, say I want to use track 5 of a certain CD as my test sequence. How does DiffMaker know where track 5 starts for each test run?

-- Bob

[bwaslo]

Bob --

Diffmaker captures two short (under a minute, preferably) recordings, both of which need to mostly be of the same material (track).  If DiffMaker is playing the track out (its "stimulus") for recording both times, then that pretty much makes it happen automatically.  But if something else is providing the source (say, a CD player), then the operator has to make sure that both recordings are mostly of the same material (and starting and ending at near the same place, within a second or two).

What DiffMaker then does (and the most difficult part to get working) is shift the two recordings (in time) relative to each other so that common parts line up.  The hard part is that this shifting has to be done to within much less than one sample (in time) to get a deep cancellation when one signal is to be subtracted from the other.  If you want to know the algorithm, it is like this:
1) do a cross-correlation of the two files (that's a DSP process, utilizing an FFT).  The result of the cross-correlation is a peak at the sample number at which the "compared" file has to shift (relative to its first sample) to line the two files up to within one sample.
2) then, the sub-sample shifting is done by trial and error, divide and conquer style:  shift a little one way, see how it goes, shift a little the other way, see which is better and start the next loop from there, looking not as far away each time.  You need a way to shift by less than a sample -- that's not a problem if the shift (i.e., "uniform delay") is performed in the frequency domain.  FFT the recording you want to shift, modify the phase result to apply the delay, then IFFT back to the time domain.  You also need a way to tell when the shifted position is better for getting the deepest cancellation over the most relevant frequencies.  That is kind of a secret for now, I'm thinking about doing an AES paper on the subject and don't want to give away all the tricks yet!  (But it's not all that fascinating unless you are a DSP person, maybe not even then).
3) a similar process is done for the gain adjustment, though of course that is easily done in either the time domain or the frequency domain.  The optimizer goes back and forth a little to optimize time offsets and gain adjustments for the best result.

A result of all that is that the process is kind of "analog-like" (read: not exact) -- if you compare a track to itself, you may find that the time shift and gain adjustments don't always come out to be right on zero, but may be some very small values.  After the optimization gets to within some amount of time or gain (you can change these settings) it will stop and give you the difference file it has found at that point.  Too much noise in either track can also fool the evaluation and make the evaluation be a little off from ideal as it tries to cancel the noise.

If the two recordings are too different, then DiffMaker will try, then stop and give you an error message.  The assumption is that most of the recordings are nearly identical to each other (otherwise, why check with DiffMaker?).

Hope that's what you wanted (or maybe too much?)

[JoshK]

Excellent.     I've been debating picking up Praxis with the audpod.  The DiffMakr software sounds fascinating.  That is a very clever way to test. 

[Bob Reynolds]

But if something else is providing the source (say, a CD player), then the operator has to make sure that both recordings are mostly of the same material (and starting and ending at near the same place, within a second or two).

That is exactly the problem I was considering how you were solving. I was not aware that DiffMaker could be the source -- that solves that issue.

Hope that's what you wanted (or maybe too much?)

Perfect! I'm a software developer so I appreciate the algorithm, but I don't understand the DSP aspects. I hope to educate myself about DSP -- its application to audio is very interesting to me.

Thanks for taking the time to be so detailed.

[miklorsmith]

I have all the hardware.  Bring it on!

[brj]

From the website:

When might you use Audio DiffMaker?  When you are interested in whether an audio signal is actually being changed by:

<much deleted>

Devices said to modify electrons or their travel, such as certain treated "clocks"

It looks like a nice piece of software.  I'll have to check it out when I get my system up and running again.  (Although to do so will finally confirm that I should make my Linux box multi-boot capable again.... grrr....)

[tgarson]

OK

So cast my vote for a Praxis Circle.....Is that all there is to it, or do I have to pull a lever somewhere?

Tom Garson
Aural Technology

[Klaus]

2) then, the sub-sample shifting is done by trial and error, divide and conquer style:  shift a little one way, see how it goes, shift a little the other way, see which is better and start the next loop from there, looking not as far away each time.  You need a way to shift by less than a sample -- that's not a problem if the shift (i.e., "uniform delay") is performed in the frequency domain.  FFT the recording you want to shift, modify the phase result to apply the delay, then IFFT back to the time domain.  You also need a way to tell when the shifted position is better for getting the deepest cancellation over the most relevant frequencies.  That is kind of a secret for now, I'm thinking about doing an AES paper on the subject and don't want to give away all the tricks yet!  (But it's not all that fascinating unless you are a DSP person, maybe not even then).

Fascinating is exactly the right word.

Best regards, Klaus

[Patb]

I vote YES for the Praxis Circle.

Pat Brown

[bwaslo]

Thanks, everyone for your "votes" and positive comments.  I have no idea what amount of posting is needed before a separate circle is allowed.  Probably enough that the traffic starts cluttering up the generic "Industry Ads" circle -- we have quite a way to go, I suspect.

On a related note -- has anyone who has tried DiffMaker made any example files or tests they'd like to share?  Or, any problems using the program? If you have a 'dyf' file to share, let me know and I can make a page for them at the LI web site.  I'd really like to hear some results of cable tests.

Bill

[jmeyer]

I vote for a Praxis circle.

Jim

[JohnR]

I'm all for it, and it's great that new members are registering to support the Liberty Instruments circle, but could they engage in the discussion at all - ?

[bwaslo]

JohnR, good point.  Discussion doesn't have to be about just PRAXIS or DiffMaker, but even about measurements or processes in general?  What would most people like to be able to demonstrate objectively, investigate, or measure?  I can't promise that I'd jump in and start programming whatever is on the wish list (I already have a pretty long backlog of requests).  But some brainstorming here might lead to something new and interesting.  And it would be interesting to just know what the technically-minded people are thinking.

On the subject of DM:
Has anyone tried the DiffMaker yet, does anyone have any results to share?  Does it seem too difficult to use, is there something you see missing in the concept?  Anything you see that should be changed or added?

If no one has results to share yet, would anyone maybe care to lend some hotshot cables or other tweaks for me to test?  I'm not willing to spend my own $$ for them (yes, I am doubtful about them having an effect, though I'd really be thrilled if I could find one that did and be able to investigate it).  I doubt dealers/mfgrs would lend me any of these kinds of things if they knew what I was up to with them and I don't want to borrow them under any false pretense).

[Aether Audio]

Bill,

You don't know me but I know about you - or I should say "Liberty Instruments."  I have been eyeing your stuff for (geez how time flies) years now and wishing I could scrape up the extra $$$ to get a system together.  I have been using TEF all along and although it has served me well, as you know there's a lot it can't do - especially compared to Praxis.  I even sent you an e-mail quite some time back inquiring as to costs, etc.  Money has always seemed to find its way into some black hole of necessity though.  But the day will come...soon I hope.  IMHO, you have a great product that is well worth every penny you ask for it and your presence here would be a feather in all our caps.

In that, I'm using my immense influence here at AC to convince the powers that be to give you your own circle.   OK...it looks like you may need more help.  For what it's worth though, you have my vote. 

Good luck,
-Bob