I equalize my system and the system wide EQ setting works the same for my AAC files as it does for my CDs. I'm not making massive EQ shifts though. I would think that if you were making enough of an EQ correction to interfere with a lossy codec, you should probably address the source of the imbalance rather than try to EQ that broad a correction. I've never heard anyone say they ran into problems EQing lossy files. I think this is something that only exists in pure theory, You'd never get that far in an extreme corner in real life.
With strong EQ, auditory masking at or around the altered frequency won't give the same result. I guess it's possible for the outcome to be audibly different, if the codec dismissed something at a quieter level that turns out audible only once the EQ reduces the overal masking at that particular freq.
Perhaps something like that?
I would think so.
Parametric equalizing use heavy processing.
The parameter there are something like:
Crisp -Thick
Detox - Vivid
But then MSEB is really well implemented, and using it in moderation is really enjoyable (somehow I never used much conventional frequency-based equalizers).
Having started it, I think I will have to do a blind test ("deaf" test, as my wife joked about it).
I will use a single .flac file - converting it to different mp3s and then asking someone to convert mp3s back to .flacs and comparing it blindly.
Any suggestions for this approach?
I have a test file set up with FLAC and three different codecs at 192, 256 and 320. It's all packaged in a single FLAC file. If you would like to use that I will send you a copy.
I equalize my system and the system wide EQ setting works the same for my AAC files as it does for my CDs. I'm not making massive EQ shifts though. I would think that if you were making enough of an EQ correction to interfere with a lossy codec, you should probably address the source of the imbalance rather than try to EQ that broad a correction. I've never heard anyone say they ran into problems EQing lossy files. I think this is something that only exists in pure theory, You'd never get that far in an extreme corner in real life.
It is hardly a problem but an observation; trying to relate it to your favourite concept of "transparency", it would be more of a test of resistance to the distortions, where more compressed mp3 start to "fall apart" by sounding less natural earlier. It happened at the extreme of the settings, which most would not use normally.
Then I also started to feel the responsibility to confirm using the same source file.
I have a test file set up with FLAC and three different codecs at 192, 256 and 320. It's all packaged in a single FLAC file. If you would like to use that I will send you a copy.
Corrections that broad are getting into the realm of mixing, not normal playback. It's a given that for mixing you want a file with a lot of headroom. It's likely that you would be bumping into the noise floor of 16 bit as much as the lossy codec.
Corrections that broad are getting into the realm of mixing, not normal playback. It's a given that for mixing you want a file with a lot of headroom. It's likely that you would be bumping into the noise floor of 16 bit as much as the lossy codec.
In those tests, it is really the distortions started to be apparent, so I would not worry about noise levels (unless I am missing something).
If .flacs are the same size and of different resolution from the previous down&up conversions that would be ideal from what I can think.
Test file PMed to you. It's all in one FLAC file and the volume is balanced. You can cut the file up if you want. Just note the order of the ten samples when you do so you can rank them in quality and send me your results. I have a list that says what each one of the samples is in the order they appear in the file.
Test file PMed to you. It's all in one FLAC file and the volume is balanced. You can cut the file up if you want. Just note the order of the ten samples when you do so you can rank them in quality and send me your results. I have a list that says what each one of the samples is in the order they appear in the file.
The test was really interesting and insightful. I really appreciate it.
Great pieces of music - choral and orchestral. The latter with the orchestral overtones and the dynamic range should ideally be a good test piece.
Then the original recordings feel very limited to me in terms of their spectral range and sound stage. I compared with several of my random flacs.
I am not sure what is exactly the reason for the truncated spectral range, the original master?
I thought one of my treble exaggerating IEMs would be instructive to use (treble murder for some electronic music recordings, it is often called). Not much to exaggerate, other than some strange recording noise in the choral music that felt quite revealing.
Other than this noise, the choral piece is definitely less sensitive to variations.
The orchestral piece, the starting solo and the climactic end felt most revealing.
Nothing was falling apart with the parametric equalizing, but the extremes of equalizing still felt more revealing.
Anyhow,
1 and 2 - not acceptable;
3- better not perfect;
4 - not acceptable;
5 - second best;
6 and 7 - more reasonable;
8 and 9 - not acceptable;
10 felt the best, perhaps after 9, I've also compared to 5 -10 felt a bit better.
I did not use any file analysis, spectral or otherwise, just my direct listening impressions.
1 and 2 - not acceptable; aac192 / frau320
3- better not perfect; frau192 (the poorest quality sample)
4 - not acceptable; lame256
5 - second best; lame320
6 and 7 - more reasonable; lame192 frau256
8 and 9 - not acceptable; Lossless / aac320 (the two highest quality samples)
10 felt the best, perhaps after 9, I've also compared to 5 -10 felt a bit better. aac256
Some real randomness there! Is this based on the ability to EQ it or just listening?
Your results are pretty typical for most people who take this test. Most people can't detect any differences consistently. A small percentage can tell up to 192, but not beyond that.
The recordings are from a high end label. Can't remember which one, but they are recent recordings. Classical music doesn't really have broad response except with cymbals and kettle drums.
1 and 2 - not acceptable; aac192 / frau320
3- better not perfect; frau192 (the poorest quality sample)
4 - not acceptable; lame256
5 - second best; lame320
6 and 7 - more reasonable; lame192 frau256
8 and 9 - not acceptable; Lossless / aac320 (the two highest quality samples)
10 felt the best, perhaps after 9, I've also compared to 5 -10 felt a bit better. aac256
Some real randomness there! Is this based on the ability to EQ it or just listening?
Your results are pretty typical for most people who take this test. Most people can't detect any differences consistently. A small percentage can tell up to 192, but not beyond that.
I'm afraid I don't remember. I should have written them down. They were chosen by a golden eared audiophile in a classical music group. They came from an HD Audio source I think.
I'm afraid I don't remember. I should have written them down. They were chosen by a golden eared audiophile in a classical music group. They came from an HD Audio source I think.
I actually think the orchestral bit was BIS. Not sure about the choral. Classical music is miked from a distance though. You don't get ultra high frequencies because they don't carry that far, and the sub bass is pretty much when the contrabass or kettle drum is used.
This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
By continuing to use this site, you are consenting to our use of cookies.
