DAC suggestion - Page 7

Good example is Bricasti M1 DAC (http://www.bricasti.com/m1.html). You do not need "golden ear" to hear the difference between several filter options in that device.

http://jn.physiology.org/content/83/6/3548.abstract

My local dealer just got this in. I have yet to listen to it. Hopefully I will be able to test it out one of these days.

Yet,

Attempts to independently reproduce these results [the "hypersonic effect"] have so far been unsuccessful.
-- http://en.wikipedia.org/wiki/Hypersonic_effect#cite_note-Colloms2006-3
[Note: Wikipedia is not an independent reference.]

Judging from this review of the subject, the case for an ultrasonic reproducing bandwidth is not proven, and 20kHz remains the practical limit for the chain as a whole.
-- Proceedings of the Institute of Acoustics, Vol. 28. Pt.8. 2006

As a result, no significant difference was found between sounds with and without very high frequency components among the sound stimuli and the subjects.
-- NHK Laboratories Note No. 486 (2004)

It's not much of a mystery. High frequencies can produce distortion in the audible range. That's now understood to be the explanation for that paper's unique findings.

If so, one wonders how those filters are distorting the signal.

Any filter distorts signal in either frequency or time domain, the difference is how much and what type of distortion it adds. If it wasn't the case we wouldn't have dozen of possible transfer functions implemented in different DACs. Every sharp cutoff filter is a compromise, those which operate in real time even more so due to limited computing power.

But if they're distorting the signal enough to be audible, their virtue escapes me. (And that if is intentional. I'm not conceding that they are in fact distorting the signal that much.) I want undistorted sound out of my source components. At least four of its five filters are doing something wrong, by that standard.

Really?

Here are the specs:

http://www.bricasti.com/images/M1.pdf

"
Table of filter characteristics at 48khz.

48khz Passband Stopband Passband ripple Stopband attenuation delay
Filter 0 20kHz 24kHz .001dB 111dB .73ms
Filter 1 20kHz 24kHz .005dB 102dB .63ms
Filter 2 19.5kHz 24kHz .004dB 103dB .56ms
Filter 3 19kHz 24kHz .005dB 102dB .51ms
Filter 4 18.5kHz 24kHz .003dB 106db .50ms
Filter 5 18kHz 24kHz .001dB 114dB .50ms
Filter 6 21.8kHz 26.3kHz .0002dB 110db .72ms
Filter 7 20kHz 24kHz .001dB 110dB .7ms
Filter 8 20kHz 24kHz .001dB 110dB .7ms

Table of filter characteristics at 44.1khz

44.1khz Passband Stopband Passband ripple Stopband attenuation delay
Filter 0 20kHz 22.05kHz .001dB 110dB 1.43ms
Filter 1 20kHz 22.05kHz .161dB 71dB .72ms
Filter 2 19.5kHz 22.05kHz .046dB 82dB .72ms
Filter 3 19kHz 22.05kHz .014dB 92dB .72ms
Filter 4 18.5kHz 22.05kHz .005dB 102db .72ms
Filter 5 18kHz 22.05kHz .001dB 112dB .72ms
Filter 6 20kHz 24kHz .0002dB 111db .78ms
Filter 7 19kHz 22kHz .001dB 110dB 1ms
Filter 8 20kHz 22kHz .001dB 110dB 1.5ms
"

All ripple is less than 0.1 dB (and not by just a little!) and all filters are flat out to at least 20 KHz.

What am I missing here?

In the previous post (189) I provide the actual specs of the filters.

All appear to be flat out to 20 KHz, with far less than 0.1 dB ripple.

AP1 appears to be inventing new science when it comes to audibility. ;-)

Wise, not to concede to AP1's apparent false claims.

Actually, all options represent brilliant performance. But there will be no end of audiophiles and reviewers who will publish a short poem about what they think it sounds like - all based on the sheerest of all imagination. If they harnessed that literary skill to writing about the partner of their choosing, at least they'd get a little lovin' ;-)

Specs are just that - specs. They are always lImIted to WHAT was measured. I heard the difference switching between filters in that device. Sometimes more, sometimes less. Instead of theorizing here go to a dealer and listen yourself. Then come back and tell about what YOU heard.

Sigh. Some people just never learn.

Typical audiophile rhetoric. The theme is that no matter how many listening tests you've done related to products like these, and differences like these, this new toy is magic and breaks all of the rules. Their sighted tests trump everything else in the world.

It is likely pure solipsism.

Logic says that the differences between the various filters are well-characterized by the vendor's spec sheet because it is highly unlikely that anything else with audible significance changed.

The variations involved are often in the hundreths and even in thousandths of a dB. At 20 KHz a 5.0 dB difference can slip by everybody's ears for reasons I've previously explained.

Arnyk, can you make it to Royal Oak/Clawson area this weekend? We can test this DAC out at David Michael Audio.

Would you recommend taking one of these to the dealer when doing that? http://images10.newegg.com/ProductImageCompressAll300/82-461-009-02.jpg

Is he equipped to do good clean DBTs?

Otherwise, not worth the trouble to even think about.

Actually, it would take more than that.

The biggies are listening to exactly the same music and getting the switching time down to subsecond and DB. Otherwise the evaluation is hopelessly compromised.

Besides, I'm hundreds of tries down this road and it is about as much fun to do yourself as watching paint dry.

Let some golden ear actually do a good test with positive results, and maybe it would be worth the trouble to run a confirmation. ;-)

I think we could make it work. All of his gear is accessible enough to where we could have someone swapping out components within 20 - 30 seconds. You're not going to have instantaneous switching, though.

Nowhere near good enough. If the filters really did sound subtly different (which they obviously don't), that time delay would make it impossible to identify them in a true blind test.

That's hopelessly compromised in terms of sensitivity for small SQ differences. Over 30 times too much switch over delay to hear small differences. It needs to be well under 1 second.

The only beneficiaries of this test if done right is going to be someone other than me.

You can pay me to deal with this sort of thing, time and materials at $120/hr with a 4 hour minimum. I can refer you to someone in the area who is better known than I to do it for a higher price. ;-)

Nobody wants to know the inconvenient truth at that price!

Never mind that. He's a high-end dealer. Do you think he's gonna cooperate?

I have a very good relationship with this dealer. I would be able to work it out. But alas, Arnyk must be paid for his time and nobody as small time as myself is going to be able to pay Arnyk for his scientific research.

I do not think you appreciate the attitude that such dealers have about blind testing. It is a mortal threat to their economic survival, and they know it.

My suggestion is that you buy the unit, and then offer to take it to Arny's for the test.
Edited by mcnarus - 8/14/12 at 9:31am

They already have a room setup with room treatments and such so I think it would be a good location to do the test. But, for that matter I would suspect Arny has a treated room as well.

I will see if I can borrow the DAC for a bit.

I do just about all of my technical work either on-site or with headphones. My listening room is untreated.

You do not need this. In Bricasti filter algorithm is switched with simple push of the button. Everything else stays the same.

Yea this is correct. I was thinking that we were testing different DACs for some reason but the filters would be changed from the remote. So in that case level matching and such would not be required. Give someone else the remote and have them change the filter for you.

One way to do blind testing with devices like this is to have a person who is hidden from all of the listeners.

Let him flip a coin in secret and make a numbered secret list of unknowns. Since there are more than 2 choices he should flip a number of coins, count heads as 0 and tails as 1 on the first coin he picks up after each toss, then heads as 2 and tails as 3 on the second coin he picks up, etc., and sum the value of the coins tossed for each unknown.

Then when the listeners call out a trial number, he consults the list and selects the filter associated with that number. They write down the filter number they think they are listening to and keep their results to themselves.

When all of the trials are complete, compare the secret operator's list with the listener's lists.

Score in accordance with this:

http://home.provide.net/~djcarlst/abx_bino.htm

What headphones do you prefer?

Nonetheless, you should confirm this with an onsite measurement. Lots of audio components do not behave as advertised.