Headphone and bitrate test

[chrisfromalbany]

so need to find out if 30 seconds is ok then.. need to come up with some 30 second clips of music. The fun part. Thoughts of Tool - 46&2 drum solo and Phathom of the A - point of no return(end of the song). I mean 30 second clip from dark side of the moon. (The remaster 1984 version also comes to mind)

 

[dmb367]

Quote:

Originally Posted by Silfer If I'm right then someone could make some clips, encode them in different bitrates and convert them to wav again and upload those files so every headfier can download them and we'll get much broader results. I know a simular test has been done a couple of years ago on another board but that wasn't a hi-fi community. I for one would be interested to see the results of head-fi listeners!

Second this idea, but is 30 seconds enough to get a good feel for a song? I mean, you're looking for some variation in the music to really get a feel for the enconding. I think, unless it was a f*ing crazy 30 seconds, that alot would sound similar in such a short time period.

 

[chrisfromalbany]

Well I think you have to choose the 30 seconds wisely. Maybe it could even get a collection of moments as 30 of only one song. What I do to compare is listen to 5 to 10 seconds of a track and then skip to the other track and listen to 5 to 10 seconds.

It kind of reminds me of an ear exam.

track #1 or track #2..
Here is track #1 again..
here is #2..
ok..
Now Track #2 or Track #5..
ect..

But hey that is just my way of testing tracks.. I listen to a few moments and compare a few moments rather then the whole song. I tend to get lost in the song or actually enjoy listening to the song rather then judging the quality.

 

[dmb367]

Thats a great point, I didn't notice but I do the same.

I got lost imagining 30 seconds of an intro or something then, boom the song's over!!

 

[jefemeister]

Quote:

Originally Posted by Silfer here It's easy to find your own threads by going to your profile and click on 'find all threads started by jefemeister'

thanks. I forgot that I started that thread. I was thinking I just responded to someone else's.

Anyway, the reasons for actually shipping out CDs were 2-fold. One, because everyone would get a different disc with the tracks in a unique ordering and no other identifiable characteristics. This would cut down on "cheating." And two, I think 30 seconds isn't really enough time. I prefer people to hear the entire songs. This experiment really couldn't be carried out in A/B fashion since it might be all but impossible to normalize volumes between tracks. Your brain will always perceive the louder track to be the superior one in A/B tests. I didn't really think about copyright isssues though.

A good point was brought up in my other thread: it probably makes the most sense to burn the music as full-resolution data tracks as opposed to standard CDs. I can create the encoded tracks myself now, as I've learned quite a bit about all of this in the last few months.

I will design the experiment and then post it here. Might take a little bit though as I am very busy with multiple job commitments and just moved. I'll keep you all posted...

 

[taymat]

Quote:

Originally Posted by chrisfromalbany Well I think you have to choose the 30 seconds wisely. Maybe it could even get a collection of moments as 30 of only one song. What I do to compare is listen to 5 to 10 seconds of a track and then skip to the other track and listen to 5 to 10 seconds. It kind of reminds me of an ear exam. track #1 or track #2.. Here is track #1 again.. here is #2.. ok.. Now Track #2 or Track #5.. ect.. But hey that is just my way of testing tracks.. I listen to a few moments and compare a few moments rather then the whole song. I tend to get lost in the song or actually enjoy listening to the song rather then judging the quality.

This is how I compared lossless/320/alt preset xtreme/standard and 128 mp3s and it's the easiest/best way in my opinion. I can hear the smudging of fine detail, metallic treble and echoey vocals at 128. At alt preset standard up it becomes 50/50 for me. I'd say perhaps I heard a tiny difference between standard/xtreme but I'm not sure. There's certainly no audible difference between lossless/320/alt preset xtreme to my ears.

 

[HiFiRE]

No one mentioned hydrogenaudio?

Link.

 

[wolfen68]

Quote:

Originally Posted by jefemeister thanks. I forgot that I started that thread. I was thinking I just responded to someone else's. Anyway, the reasons for actually shipping out CDs were 2-fold. One, because everyone would get a different disc with the tracks in a unique ordering and no other identifiable characteristics. This would cut down on "cheating." And two, I think 30 seconds isn't really enough time. I prefer people to hear the entire songs. This experiment really couldn't be carried out in A/B fashion since it might be all but impossible to normalize volumes between tracks. Your brain will always perceive the louder track to be the superior one in A/B tests. I didn't really think about copyright isssues though. A good point was brought up in my other thread: it probably makes the most sense to burn the music as full-resolution data tracks as opposed to standard CDs. I can create the encoded tracks myself now, as I've learned quite a bit about all of this in the last few months. I will design the experiment and then post it here. Might take a little bit though as I am very busy with multiple job commitments and just moved. I'll keep you all posted...

Before you dive in, maybe we could have some forum consensus on the experiment design (let the group do the work!) then we can figure out how best to make it happen. This might be getting out there, but to make it fun, have a friendly competition in which one of the people correctly picking the formats/bitrates is drawn from a pool to win a modest price. I would even donate a few bucks towards that

.

Is a separate thread required to explain the project and solicit feedback. If so, who would like to post that?

 

[ILikeMusic]

This is the sort of thing they do on Hydrogenaudio as a standard procedure, and it almost always shows LAME -alt preset standard and above to be virtually transparent to the source (and the guys over there are very trained listeners when it comes to compression artifacts and such.) Unfortunately, accurate as that assessment is I doubt it will ever be accepted here.