0.9y-7 and Crack Detect criteria with KS and DAP

[Calibrator]

hey there Peter,

did you ever get around to revamping the trip conditions that bring forth the crack detect message when using KS with DAP ?

Just queued up Stanley Clarke's "1,2, To The Bass" album , track 1, and I get a ...

"24 bit second stage crack detect -> 494"

followed by ...

"44100/4/176400/62640816/50000000"

I recall reading about your intention to look into this but can't recall reading if you did in any of the recent release notes.

No issues if I disable DAP.

Can make that track available if you need for testing.

It is a rather dynamic album too, and that first track would wake the dead when it starts ... haha

Cheers,

Russ ( the one with a permanent smile everytime XXHE fires up  )

[PeterSt]

Dear Russ,

But I did ! I forgot to mention it in the Release Notes though. However ...

I highered the tripping point from 20 to 200, and you now have 494.
I guess I will higher it again, but at some stage it will get dangerous. On the other hand it must have been ages ago by now that someone discovered a real problem by means of this.

I just set it to 1000 ...

Peter

[PeterSt]

I just set it to 1000 ...

FYI : Which only happens when Arc Prediction is used. Without that it's still at 20.

[Calibrator]

I just set it to 1000 ...

Thanks Peter.

Can I presume that the number you specify has some bearing on the digital value one sample point can vary from the previous? This would make sense as a small number would indicate a gentler slope ( allowable ) than a bigger number.

I suppose I could load the track into Sound Forge ( or similar ) and zoom right down to a per sample view and look at the waveform shape. Might give me an idea, however after your AP algorithm has done its thing it's quite likely the newly created "intermediate" sample points won't look like a simple average between the originals.

Ta again,

ciao

[PeterSt]

Can I presume that the number you specify has some bearing on the digital value one sample point can vary from the previous?

It is about the number of times a by me defined (fixed) transient is happening, such a transient (by me) appreciated as not much normal. The problem though is, this is derived from red book, while the moment I look at it is in the "hires" domain (if you're implying that). So, this needs a translation of what's appreciated as "can happen" in red book to the same for hires, while all is on the dB scale and I just "look" at the numbers and picked a value which seemed good to me. IOW, the problem will be more in that value chosen, than the number of times it "overs" that value. Now, without Arc Prediction all was ok (and I still have one (Turtle Records) audiophile album which trips on the 20 because it just is so), but with Arc Prediction in the middle I don't know how to make the translation. Why ? well, because with red book as the base there is no Arc Prediction. And also to remember, Arc Prediction is not something of which I will know what the result will be (for wave shapes expressed in (relative) numbers). It completely depends on the (music) content.

Now you know.
Not.

[Calibrator]

Thanks Peter,

I think I understand where you're coming from. You are trapping a trend of abnormality over a predetermined timespan ( albeit small ) rather than a digital difference between adjacent sample points. A sudden large jump between 2 points ( if that was all there was ) would probably go unnoticed, whereby if that jump were to continue over a dozen or so ( or more ), might.

Thanks teacher 

Russ

[Calibrator]

Happy to report 0.9y-8 plays that album just fine now. 

Russ ( enjoying a little Pavarotti while partaking of morning tea in sunny Melbourne  )