[stv014]

I plan to create some ABX tests, and am looking for good samples that could be used. If anyone is interested, post your suggestions, or even better links to actual files (of course, without copyright problems, if possible). It would probably be best to use multiple short (<30s) samples from various types of music, with different dynamic range etc. Limiting the length is needed also for copyright reasons. A good collection of sample files could be useful for future tests, too.

 

[Speedskater]

Depends on which question you want to ask:
a] Is it possible to hear the slightest difference between A & B?  That would be some nasty test tones or signals and stereo applause or a few brief parts of musical selections.
b] is it possible to hear some differences when listening to real world music?
 
Remember that an ABX test can detect very small differences.  Way to small to form a preference.

 

[stv014]

My idea was to use a few short parts (e.g. 5 samples of 20-25 seconds each) from a selection of various music. Synthetic test signals are indeed the best for revealing specific artifacts they are designed to make the most audible, however, I suspect those who cannot hear a difference would use the argument that "it would have been audible with real, complex music". Although creating the tests is probably not worth spending the time on after all, the results would not be accepted anyway, and there is not much interest either.

 

[xnor]

Check out the samples used for listening test at the HA forum. They usually use everything from pure speech to classical, pop, .. metal.

 

[stv014]

Thanks, checking the hydrogenaudio.org forum is indeed a good idea, since there are many ABX related threads there. In fact, I already had a look earlier, but did not spend too much time searching for samples.

 

[Speedskater]

You can never prove to others that the units sound the same, it's a one-way street.
The real question should be:
Just how much money is that very small sound difference worth to you?

 

[xnor]

Send IgorC a PM, he did the last test afaik. He'd probably even send you all the samples used if you ask kindly.

 

[stv014]

Quote:

You can never prove to others that the units sound the same, it's a one-way street.

 
That is true, it would be most useful to those who are not convinced yet either way, but see a lot of claims that there is a large difference, and others that say there is none, without having any way to verify them. Of course, no test can ever prove with 100% certainty that there is no difference at all, but it is possible to get an idea whether there is a significant difference.

 

[bigshot]

You don't say what you're comparing... Redbook vs high bitrate, lossy vs lossless, etc.

There are a few recordings I use to test... Reiner's Living Stereo version of Tchaikovsky's Marche Slav (good for testing subtle bass differences), Fiedler's Living Stereo version of Offenbach's Gaietie Parisienne (good for overall response balance and dynamics), Munch's Living Stereo version of Ravel's Daphnis et Chloe (good for overall response and complex textures in very subtle gradations), Dorati's Decca version of Stravinsky's Rite of Spring (dynamic punch) and Sammy Davis Jr's Decca Sides CD (extremely difficult to compress without artifacting).

With these five recordings, I can fine tune anything.

 

[JMS]

According to Harman's blind testing, #1 most revealing track is pink noise, and close behind is Tracy Chapman's "Fast Car".

 

[anetode]

By the same token as Chapman's song, the killer clip during the development of the mp3 codec was Suzanne Vega's "Tom's Diner".

 

[budget minded]

well, i don't know about "sound samples", but some good test tracks i like to use are:
water drums 1 - baka forest people... it's a really good test of imaging as you can hear the water swirling as it's drummed. it's also a good test of EQ too. it sounded a lot warmer on my old superzeros than airier energy RC-10s. it's available on a great 2 CD rykodisc collection called "the best of both worlds" with a wide variety of amazing songs with many test worthy tracks including one from mickey hart's planet drum which used to be considered a good demo disc and just hearing it recently on my new aluminum domes, it's really great for testing air besides percussion, imaging & bass
the tide is high - blondie... great test for female vocals, percussion & imaging
almost ANYTHING by buzzcocks. they record their stuff very bright, and it can be very revealing on how forgiving a particular speaker is. they cansound harsh on lower quality speakers where the brightness is less of an issue with better resolved gear
gravikords, whirlies and pyrophones is a fun CD with all kinds of home made instruments with a few really good percussion tracks and some great imaging tracks too. like the rykodisc compilation, much of it is well recorded
carta a l'exili - badi assad has a REALLY high frequency flute sound that is ruthlessly revealing of harsh tweeters. it was the only track that i can recall that stressed my superzeros otherwise polite sound and surprisingly sounds a lot purer on my energys. the entire CD is well recorded and has been called one of the highest fidelity recordings of jazz guitar. i don't like jazz, but i love that CD.
dark side of the moon is a classic demo disc with bass, imaging, percussion & vocals to demo off.
finally
stronger than pride - sade is a nice test of vocals. it was a really good track to show how much it cleared up my superzero's midrange & treble over my too warm NAD that might synergize better with my also lean energys.
i don't know why - norah jones is supposed to be a great vocal demo track. i could have sworn i already had it, but couldn't find the CD to test my new speakers. it's also available in high rez
woofer cooker - bass hit, as far as i know, is "the" track to test how low a sub can go with some infrasonics
 
those are all tracks i burned on a CD i take with me any time i'm looking at any hifi gear, not only to reveal what it does right, but things it can do wrong too.
 
you can also download some free high rez demo tracks at various sites that sell downloads.

 

[stv014]

Quote:

Send IgorC a PM, he did the last test afaik. He'd probably even send you all the samples used if you ask kindly.

 
For 44.1 kHz/16-bit tests, IgorC's LAME_MP3.zip (20 FLAC samples) seems to be suitable, thanks for the suggestion.

 

[stv014]

I created a ~3 minute sample from the above mentioned package. I tried to include a reasonable variety of different sounds, although the choices could probably be better.