You're wrong, 5uS fixed delay in one channel, not jitter, is not something anybody can hear, despite your claim that you can.... It's the same as moving one speaker about 1.7 mm....
Please understand that the dam1021 will never get the option of synchronous interconnect of multiple boards, it's just not possible without major hardware changes. Why try that when I already offer the dam1121 and dam1941 ?
I posted the link to test tracks on HA and I did the ABX test myself resulting in higher than 95% confidence. I hope you can at least accept that result and realize that the problem with async connection is indeed audible at least in controlled environments. You also haven’t answered my very simple question of whether a higher input sampling rate would potentially reduce the delay. I would not have repeated myself if you had responded in a respectful manner.
If the problem is hardware, I do have to agree that the only solution is to move on to a newer product line, given that the audible uS delay poses enough problem for one to justify such. I only thought a software solution possible because of your post in 2015 and discussions on the potential uses of FPGA SLV and MCLK. If this is not the case and synchronous operation is impossible in the current state of hardware, then we can only accept as it is the random delay between channels every time the system relocks to a signal. I hope you can understand my concern.
—————
I read your posts from 2015 again and I cannot but believe that a software solution is indeed possible, as you have suggested yourself 3 years ago, though the development will take some efforts. I believe there are very good reasons to offer this firmware function but it is utimately up to you since the firmware is proprietary. You might not care enough about improving an old DIY line product, you might not care how we think about it, and it probably won’t affect your sales, but we will think what we will. I’m sad to find out that the standards here are lower than I thought, but I’m content with an understanding of the situation.
Last edited:
A couple of times recently I have said:
async boards will follow each other down to less than one sample.
In most cases, I'm not going to add new features to old products.
And yes, over time many things have been discussed.
One sample at 44,1k is about 22uS if I did the math right, so how could we even test 5uS at 44,1k samplerate?
I can assure you that you won’t even hear one sample deviation from one speaker to another.
Things might be different, if you mix otherwise identical signals with one sample offset, but that is a different case.
Agreed, but to move a file to create an offset, it has to be in sample increments. If that was done at a higher sample rate, 5uS are possible, but I would guess the following downsampling introduces its own variables, and you might hear something else than pure timing differences.
Just a guess, though.
Ynmichael, can you post the link to the test file again, I cannot find it.
It was an earlier test from hydrogen audio... I tested 20uS 10uS and 5uS all detectable in ABX with very high confidence.
It’s in this thread: Audibility of phase shifts and time delays - Page 2
What I heard what clearly a positioning difference. Hi res would probably strengthen the effect
Quote: If you edit these files and zoom in on one of the impulses and compare the L&R channel you should see the delay which will be less than 1 sample wide for the last 3 files. This should put to rest any misapprehensions about 44 Khz sampling not being able to encode time differences with high resolution.
———————
So to make clear what Soren wouldn’t: dual-mono in 1021 will typically have less than 22uS delay in 44.1Khz mode, and by extension, less than around 2.7uS with external x8 oversampling. If synchronization before FIFO were available, no further jitter would be added and the delay may be reduced by a factor of 16 due to bit clock tracking. If it’s after FIFO, there would be 0 delay but tiny amount of jitter added due to extra clock wiring.
Last edited:
This test would be valid if this step did not exist: (5) All files were down sampled to 4416
One simply cannot distinguish temporal differences of 5uS with a SR of 44.1. There may be something that you are hearing, but it is an artifact of the processing & conversions and not an actual temporal difference.
One simply cannot distinguish temporal differences of 5uS with a SR of 44.1. There may be something that you are hearing, but it is an artifact of the processing & conversions and not an actual temporal difference.
With a perfect reconstruction filter it would be valid, but that does not exist.
I meant given different input sampling rate, output should be indistinguishable, if the resampler is proper.
We discuss all the different resampling options for a reason, none is proper enough and we look for the least destructive...
Confusing. Excactly... what? ... either... you dint make sense.
Firing from the hip?
//
Linear filter with a large number of taps is pretty much perfect in my opinion. You can prefer other sounds though. And btw 4K soft has slightly lower THD, and NOS the lowest.
It’s not a perfect argument I guess. I still can’t help but believe that 5uS and probably even 2.5uS is perfectly audible in ABX though. Even if it doesn’t take away from musical enjoyments (do we worry about a slightly off center seat in the concert hall?), we should be aware that the dual-mono setup is not reference quality at all in some aspects for any discerning person.
Nor is it suitable for scientific or measurement purposes, but that was clear from the start.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.