Skip to main content
Topic: Designing a good listening test (Read 504 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Designing a good listening test

So I ask the experts, please help me design a good listening test. There will be no instant switching unfortunately.

I just need someone to walk me through the steps. I may be in a position to compare my AVR in my bedroom to a friends Mark Levinson Preamp and power amp combination. I'm curious to see how much difference there will be, but I need your guidance in order to construct the test.

Any pointers or advise would be most appreciated.

Re: Designing a good listening test

Reply #1
How "scientific"?

See ABX test.  That's the ideal.  

For a casual non-blind A/B test, do your best to level-match.   If you have a multimeter to measure the amplifier output voltage into the speakers, that's ideal.   Or, you can use an SPL meter.   Otherwise you can set-up a laptop, or a phone with a recording application or anything that has a meter.   The voltage or SPL level isn't important as long as both amplifiers are level-matched.  If you measure acoustically, use a lower frequency for more-reliable more-consistent readings.     I was doing some high frequency measurements (~8kHz) awhile back, and just walking around behind the SPL meter changed the readings.  Audacity can generate test-tones for level matching. 

Of course, make sure tone-controls/EQ are set to flat.    With the AVR, you'll have to make sure any automatic room-adjustment or subwoofer crossover disabled.

If you want to approximate a blind ABX test, you'll need to randomize the selections (flip  a coin) and you should disconnect & re-connect the speakers even when comparing "A" to "A".     With most amplifiers you can connect/disconnect the speakers with power-on, but it's a good idea to do it with no signal and be careful not to short-out the connections.   If you can rig-up some kind of speaker connectors that you can easily un-plug and plug-in, that's better than switching wires to binding posts, especially with "live" connections. 

Get some Y-Adapters so you can feed the signal into both amps continuously without switching.

Since it won't be a double-blind it's best if the person doing the switching is hidden from sight along with the amplifiers.

There might be a pop/click when you connect the speakers and that could be a give-away and if you turn the equipment on/off there could also be give-away clicks or thumps, etc.

Quote
There will be no instant switching unfortunately.
It's easier to hear a difference with instant switching, but in the real world you wouldn't be switching amplifiers that fast.



Re: Designing a good listening test

Reply #2
please help me design a good listening test
If one was a undercover guerilla peddler of audio jewelry, one might suggest something like this: http://www.diyaudio.com/forums/solid-state/194882-double-blind-auditions-thread-6.html#post2679256 rather than a mentally taxing ABX.
Thereby making it impossible for the average Joe, so that casual sighted listening is reverted to as the more attenable possibility in a convoluted false equivalency.
Oh wait, did you say no instant switch? Maybe a dart board and darts?
Or there is always a good fishing rod.
Loudspeaker manufacturer

Re: Designing a good listening test

Reply #3
please help me design a good listening test
If one was a undercover guerilla peddler of audio jewelry, one might suggest something like this: http://www.diyaudio.com/forums/solid-state/194882-double-blind-auditions-thread-6.html#post2679256 rather than a mentally taxing ABX.
Thereby making it impossible for the average Joe, so that casual sighted listening is reverted to as the more attenable possibility in a convoluted false equivalency.
Oh wait, did you say no instant switch? Maybe a dart board and darts?
Or there is always a good fishing rod.

That link is a hoot. Why are they claiming ABX test have a strong bias against positive results?

Re: Designing a good listening test

Reply #4
That link is a hoot. Why are they claiming ABX test have a strong bias against positive results?
Well, this is about as close as our undercover guerilla jewelry peddler has come to confessing his real motive: http://www.diyaudio.com/forums/the-lounge/325234-evidence-consider-sufficient-90.html#post5523325 "ties" to "audio development".
Now his co-conspirator JKenny and fellow guerilla peddler of magic dac fame, is posting under the mmerrill99 pseudonym due to some previous difficulties on that site, can't admit to such agendas due to risk of perjury trap.
The Dunning-Kruger/Shyster "logic" goes something like this:
ABX is "imperfect", due to "stress" and cognitive load, "masking" real "differences", so unless any blind/controlled test done adheres to cherry picked ITU standards impossible for the average Joe at home, the results are useless. Therefore, they are now equivalent to totally sighted, uncontrolled "listening". So anyone considering buying an organic Irish DAC or some shyster German bling, should consider them equivalent. Obviously the sighted/uncontrolled listening is the easier, more relaxing option, so go with that.
Now, when asked exactly how these "hidden" differences were identified originally, to know that they were "masked" by ABX, or when asked what these rumored "positive controls" used for the ITU standard tests of magic bling are, expect only posterior derived obfuscation.
Expect references to tests that used no ITU standards, no "positive controls", but seemed to yield the desired results, "positives", rather than the dreaded "false negatives" that plague controlled testing of magic bling.
You can't make this stuff up.
Btw, both were members here once, but obviously that didn't go too well, so they sought greener Dunning-Kruger pastures
Loudspeaker manufacturer

 

Re: Designing a good listening test

Reply #5
thehey95,

If you decide to do a casual (non-blind) listening test and you and/or friend(s) hear a difference, please DO NOT report the results here.  (It would be a violation of TOS #8.)

 
SimplePortal 1.0.0 RC1 © 2008-2019