Axiom Home Page
Posted By: J. B. blind testing of speakers - 11/08/11 09:27 PM
can someone at Axiom tell me within how many dB's 2 different speakers are adjusted for equal volume before doing critical blind or double blind listening tests?
Posted By: SirQuack Re: blind testing of speakers - 11/08/11 10:08 PM
I believe they use a special switcher so the speakers are exactly set the same, plus other criteria. Keep in mind, placment will play a huge roll.
Posted By: RickF Re: blind testing of speakers - 11/08/11 10:19 PM
Originally Posted By: SirQuack
Keep in mind, placment will play a huge roll.

Would that be like a big 'ol drum roll Randy? confused


grin
Posted By: Ian Re: blind testing of speakers - 11/08/11 10:23 PM
J.B.

It is very important to get the SPL of each one exact. With careful set up you can get them within fractions of a decibel. As SirQuack mentioned the positioning can also play a roll so if the results are close the positions need to be exchanged and the test repeated to ensure your results do not change.
Posted By: CatBrat Re: blind testing of speakers - 11/08/11 10:24 PM
Originally Posted By: RickF
Originally Posted By: SirQuack
Keep in mind, placment will play a huge roll.

Would that be like a big 'ol drum roll Randy? confused


grin


Myself, I prefer lots of butter on mine. (I've been on a strict diet. How could I not think of food.)
Posted By: RickF Re: blind testing of speakers - 11/08/11 10:39 PM
Originally Posted By: Ian
J.B.

It is very important to get the SPL of each one exact. With careful set up you can get them within fractions of a decibel. As SirQuack mentioned the positioning can also play a roll so if the results are close the positions need to be exchanged and the test repeated to ensure your results do not change.




laugh laugh
Posted By: SirQuack Re: blind testing of speakers - 11/08/11 10:42 PM

Posted By: JohnK Re: blind testing of speakers - 11/09/11 02:57 AM
Jacques, the "roll" that using different frequencies plays when calibrating for equal speaker volume is also interesting. Primarily electronic components such as receivers and players have so little variation in frequency response across the 20Hz-20KHz range that using one frequency(e.g. 1000Hz)to set levels equal will result in audibly equal levels for the entire range. Speakers, on the other hand, have much larger variations which often occur at different frequencies for different speakers. For example, if speaker A was 2dB louder at 1000Hz than was speaker B, lowering A by 2dB would result in equal volumes at that frequency, but if B was 2dB louder than A at say 2000Hz(certainly a common possibility)then the 2dB adjustment previously made because of the 1000Hz measurement would result in 2000Hz now being 4dB louder in favor of B.

Using pink noise, which contains the entire frequency range in a way roughly similar to music, rather than a specific frequency when setting levels can result in a better average leveling result when comparing speakers with music selections. It would be interesting to hear Ian's comment on what Axiom uses in equalizing levels.
Posted By: nickbuol Re: blind testing of speakers - 11/09/11 04:35 AM
Huge roll?


or Huge Roll?


Yes, that is a roll of toilet paper.

Sorry JB.

Posted By: J. B. Re: blind testing of speakers - 11/09/11 01:15 PM
Originally Posted By: Ian
J.B.

It is very important to get the SPL of each one exact. With careful set up you can get them within fractions of a decibel. As SirQuack mentioned the positioning can also play a roll so if the results are close the positions need to be exchanged and the test repeated to ensure your results do not change.


what i would like to know is what fraction of a decibel are those speakers usually adjusted to.
Posted By: J. B. Re: blind testing of speakers - 11/09/11 01:20 PM
Originally Posted By: JohnK
Jacques, the "roll" that using different frequencies plays when calibrating for equal speaker volume is also interesting. Primarily electronic components such as receivers and players have so little variation in frequency response across the 20Hz-20KHz range that using one frequency(e.g. 1000Hz)to set levels equal will result in audibly equal levels for the entire range. Speakers, on the other hand, have much larger variations which often occur at different frequencies for different speakers. For example, if speaker A was 2dB louder at 1000Hz than was speaker B, lowering A by 2dB would result in equal volumes at that frequency, but if B was 2dB louder than A at say 2000Hz(certainly a common possibility)then the 2dB adjustment previously made because of the 1000Hz measurement would result in 2000Hz now being 4dB louder in favor of B.

Using pink noise, which contains the entire frequency range in a way roughly similar to music, rather than a specific frequency when setting levels can result in a better average leveling result when comparing speakers with music selections. It would be interesting to hear Ian's comment on what Axiom uses in equalizing levels.


my guess is they usually use band limited pink noise, maybe something like 300-3000 Hz. this should give a good match for the average SLP put out by the speakers.
Posted By: J. B. Re: blind testing of speakers - 11/09/11 01:22 PM
there is "some" difference between the two terms: roll and role. at least i hope so. ;-)
Posted By: alan Re: blind testing of speakers - 11/09/11 03:48 PM
Hi J.B,

We try to get the two speakers within about 1/4 dB of each other, because even though our hearing cannot detect that small a difference in loudness (1 dB is considered the detectable threshold of loudness differences) tests have shown that if there is a bit less than 1 dB loudness difference, although we can't "hear" it, the itsy-bitsy louder speaker will be perceived as "better" in the blind tests.

Axiom's new listening room, still in construction, will contain a speaker shuffler to exchange speaker positions to randomize the role (that's ROLE) that room placement variations introduce to speaker testing.

Regards,
Alan
Posted By: Murph Re: blind testing of speakers - 11/09/11 03:54 PM
1 db represents the difference in loudness that an average human can detect as being different. However, does perception of sound quality (verses loudness) dip into a fractional level? This I do not know, thus I ask.

Also, it will require more than your average radio shack db meter to accurately measure into small fractions of a db and there is also the variations of the frequencies to consider as noted above.

Therefore, for me, I think that for home listening tests, getting within a db is a reasonable goal but I guess it depends on the answer to the above question.

EDIT: My post was being written as Alan posted his ahead of me. I'll let it stand as I'm too lazy to rethink my questions.
Posted By: Murph Re: blind testing of speakers - 11/09/11 04:02 PM
Originally Posted By: alan
.
Axiom's new listening room, still in construction, will contain a speaker shuffler to exchange speaker positions to randomize the role (that's ROLE) that room placement variations introduce to speaker testing.


I think that, for fun, they should invite Alan to a listening test and have a set of speakers continually 'shuffled' while he is actually listening to them. Just to see what the continual movement effect does to his comments on sound stage and discover if he can actually detect what is happening or will he end up in a fetal position on the floor.
Posted By: alan Re: blind testing of speakers - 11/09/11 05:13 PM
Aarrrgh!

Going back to your earlier question, the research shows that in the case of loudness differences of less than 1 dB between two speakers that are NOT detectable by listening (but which are measureable with a broadband pink-noise test signal), the louder of the two speakers will be perceived as sounding "better". I don't recall details of how that was defined.

At the NRC, we also found out that if you have a cold or sinus congestion, the reliability and consistency of blind speaker ratings goes out the window. The same can be said of using alcohol or weed before listening tests.

In the latter case, subjects became much less critical and even piece-of-crap speakers were highly rated, simply because the music simply sounded so great while the subjects were under the influence. . .

Cheers,
Alan
Posted By: dakkon Re: blind testing of speakers - 11/09/11 05:21 PM
Originally Posted By: alan
Aarrrgh!

or weed


Cheers,
Alan



heheheh, i found that part very funny Alan.. laugh
Posted By: Murph Re: blind testing of speakers - 11/09/11 05:28 PM
Thanks for answering my question Alan. It's always amazing how sensitive human senses can be and also how easily the senses can be influenced.

Also for letting me know that I want a job at the NRC. [Shuffles off to fake a new resume]
Posted By: MarkSJohnson Re: blind testing of speakers - 11/09/11 05:38 PM
So, what you're saying is that if we can find a reliable drug dealer, we don't need Axioms? grin
Posted By: RickF Re: blind testing of speakers - 11/09/11 05:43 PM
Alan, you have forevermore distorted my visual image of the NRC sounds labs.




laugh cool
Posted By: J. B. Re: blind testing of speakers - 11/09/11 05:58 PM
thanks Alan for your answer; and the same to everyone else too who contributed, including the good laughs. :-)
Posted By: CatBrat Re: blind testing of speakers - 11/09/11 06:45 PM

Posted By: SirQuack Re: blind testing of speakers - 11/09/11 07:11 PM
Wow, that is a very old pic of Alan, Rick.
Posted By: fredk Re: blind testing of speakers - 11/10/11 01:20 AM
Originally Posted By: alan
Aarrrgh!

Going back to your earlier question, the research shows that in the case of loudness differences of less than 1 dB between two speakers that are NOT detectable by listening (but which are measureable with a broadband pink-noise test signal), the louder of the two speakers will be perceived as sounding "better". I don't recall details of how that was defined.

[pedant]
So, we can hear differences of less than one db, it just doesn't register as louder.
[/pedant]
Posted By: JohnK Re: blind testing of speakers - 11/10/11 02:11 AM
You got it, Fred. Small differences in sound level which aren't clearly identifiable as such can instead be described as factors such as "stronger bass", "clearer highs", "fantastic imaging", etc. This is particularly noteworthy in components such as amplifiers and players which have no capability to affect those factors when played at matched volume levels.
Posted By: Ian Re: blind testing of speakers - 11/10/11 02:17 AM
fredk

The perceptibility of loudness, and other differences caused by one being louder than the other, is an interesting test. By simply having the only difference in your double blind listen test between position A and B a difference in dB you can judge the effect of this phenomenon in isolation. When you are 2 dB louder you will get comments referring to the difference in SPL along with other comments about the sound difference which will be favourable for the slightly louder one. As you get down to 1 dB the comments in regards to SPL will start disappearing leaving just the odd comment about minute advantages for the slightly louder one. Once you are down to a ½ dB difference the comments about differences of any nature will disappear. If you can keep your test products within a ¼ dB of each other then you know this variable is out of your results.

For setting up the levels we use IEC 268-5 weighted noise. It is rolled off below 40 Hz and above 5 kHz. You do want a good meter for setting this up too. Though the SPL meter only needs to read a difference between the two products and not any sort of absolute SPL, it does need to be linear and broadband. We use our B&K SPL meter as the benchmark.
Posted By: JohnK Re: blind testing of speakers - 11/10/11 02:21 AM
Ian, thanks for detailing the sound source used for calibrations.
Posted By: fredk Re: blind testing of speakers - 11/10/11 02:54 AM
Thanks Ian and John. Ian. In my mind, your posts reconciles some of the difference of opinion between different researchers. I believe it was Toole that suggested that you could hear differences as little as 1/2 db. I think it was something quoted by a third party and no explanation was given as to what difference you might hear.

For all that, my M80s sound great at all db levels up to the 'you gonna break your eardrums' setting on my receiver. grin
Posted By: MarkSJohnson Re: blind testing of speakers - 11/10/11 11:48 AM
Ian, thanks for taking the time to post this. Interesting stuff!
© Axiom Message Boards