I found it very difficult to distinguish these samples, playing the mp3s through my on-motherboard sound card into a pair of inexpensive (but not crap) IEMs. First I listened to apple, which sounded pretty good, although the level was a little hot. Then banana, which was noticeably louder, causing additional distortion on the violin piece, so I turned the volume down and started again. After that banana was still noticeably a little hotter, which made me think it was the passthrough; it was the only one that really stood out for me, so I voted for it. I found reasons to dislike damson and kiwi, but would not bet that I could hear a difference in a DBT. Orange did not impress me one way or the other. Peach and pear seemed very similar to me, and just fine. There was something I liked about pomegranate, although I couldn't put my finger on it, maybe something to do with dynamics. Raison made no impression on me, certainly nothing wrong stood out. Satsuma sounded good to me, I wondered for a bit if it might be the original, but again nothing to really hang my hat on. I would not risk a bet on being able to distinguish any of these from another, except banana where I could barely detect the level difference.
It may indicate to poor quality of the "direct" sound used for the test.
If there is any similar test in future, I would suggest to use wav files for comparisons, exactly matched in level.
The people who are testing the samples should not use cheap PC motherboard soundchips and cheap headphones or plastic speakers. The sound should pass through the best possible audio they have, wav files might be burnt to CD. It is a lot of time and lot of work, I agree, but it is needed to get some credible results.
And another unfortunate aspect: I've looked at the Master version, and the path was run too hot, this is the 5dB difference that Pano saw, and there is quite severe clipping in places. At least the clipping was consistent, but it would have had an impact on perceived quality for some people.
Guys, I have prepared two files, 48kHz/24bit, in a very high quality, for comparison. The length of each file is about 2:20 min. They are matched in level within 0.01dB (really). Is there anyone who would be willing to organize a poll, if I placed the files somewhere? I would reveal nothing about methods used until the poll has collected enough votes to be closed. Is it possible that poll results are not disclosed to others during voting?
Well done Pano! This is how it should be done.
Mooly, when we do the test again, we definitely need to match all presentations to 0.1dB or better.
Pano, is Foobar ABX, you favoured player to decide between 2 stereo presentation pairs? I have never used it before but I was a beach bum for most of this Millenium.
But doing paired presentations for 10 items would take forever.
I wonder if we can force its use.
Can you prepare a 3rd file which could be a repeat of one of the 2 or different.
If you do repeat one of them, please juggle the numbers so there is a chance the first 2 files might be identical. Don't even hint at what we are listening too .. but red herrings are good.
IMnotsoHO, 16b/44.1kHz is plenty good enough for a test of this sort.
Instead of a poll, we can just PM Pavel with our rankings & comments. Also tell him if you want your rankings/comments to be attributed to you when he releases the results.
Last edited:
If I understand your question correctly, then the answer is yes.
An amp might have not sufficient ability to properly drive the circuit in front of it. So a "stronger" amp added in front of it can help but with compromises, such as added distortion.
So when an inherently good opamp is operated in an optimal condition it may not need a "helper" to do its job properly, and no compromise to be taken.
For example, the peach, with its positive attribute I perceived during the test, given a chance to operate better in a given circuit, might be the best sounding of all. But how to put it in the best working condition is the real challenge we faced every time we design an amplifier.
But in our test, says who banana (direct) is worse than any of the opamp??? Well one member mention that pomegranate sounded better, but that was his opinion, or the preference of the majority joining the poll.
Yes, we are talking with Mooly how to manage it. But with my files (different from those in #286)
Last edited:
I am just communicating with Mooly about the procedure
OK, here is the new thread with two 48kHz/24 bit files
http://www.diyaudio.com/forums/ever...istening-test-test-your-ears-audio-chain.html
Mooly, thank you very much for your help.
I was examining the files, looking to see about ways of comparing them using software, and this popped out: a clear example of electronics misbehaving, for no obvious reason. This was the Peach vs. Master file, the former nominally well performing ... look at the right channel, the lower of the two in each track, at time 3:11.480 - the circuitry abruptly gets fed a dose of much higher frequency content, and the waveform peaks at a much higher level than it should - clear distortion taking place ...
Last edited:
These 2 files have clear different levels. This can easily be the cause of the difference in graphics.
Never use visual clues to determine audible differences, it not good practice.
http://www.diyaudio.com/forums/ever...tout-where-you-get-decide-19.html#post3700526