Skip to main content

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: Have you ever regretted ABXing? (Read 45079 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Have you ever regretted ABXing?

Reply #25
I appreciate all of these posts. I'm still de-programming as i think of it, and the though of taking my lossless file collection and turning them into a lossy format kinda makes me want to run for the hills.  I know it's "just music" but i'm one of those people who sometimes, when i play a piece that moves me i feel like i might never have to eat again if i kept listening to it.  It "feeds" me so to speak.  In resturants and other public places, background music does not exist for me. It's always part of my foreground.

What if one does ABX testing and finds that MP3s at high bitrates are 'good enough'.  Then one gets a different amp or moves the audio system to a different room? Then you'd have to test all over again to see if it's still good enough in the new room/with the new amp/and the dog not shedding as much this time of year.

(i assume that dog fur creates a -.005 db suck-out at 80 hz when a dog is standing a meter away from the speaker  )

I always knew it was a case of diminishing returns but it is nice to know that a $20k amplifier compared to a $2k amplifier makes little to no difference, according to what i've been reading lately.
Music lover and recovering high end audiophile

Have you ever regretted ABXing?

Reply #26
(i assume that dog fur creates a -.005 db suck-out at 80 hz when a dog is standing a meter away from the speaker  )


Clearly, you need a second dog so the first one doesn't unbalance the image. :-)

Seriously, I agree that it makes little sense to convert to a lossy format. For one thing, it's been reported elsewhere here that expert listeners were able to distinguish between high BR MP-3 and lossless, so there's always the possibility that, as you surmise, you'll be able to hear the difference yourself. For another, bits are cheap these days, so why bother? I still have some MP-3's that I made back in the days when they weren't, and I still listen to them, but for new files, I stick to flac.

I see that some here have given up on audiophilia entirely after discovering that they couldn't hear some things they thought they could in ABX tests. I'm not going to second guess their choices, because I think everyone has their own reasons for doing things. But from my own perspective, the pursuit of good sound is fun and rewarding, whether one can hear the difference between two amplifiers or not. So I see blind testing more as a tool than a challenge to what I like to do. Even if amplifiers do differ only in frequency response and overload characteristics, they still differ, and other components, such as loudspeakers, differ even more. So there's plenty of room for tweaking.

Have you ever regretted ABXing?

Reply #27
Then you'd have to test all over again to see if it's still good enough in the new room/with the new amp/and the dog not shedding as much this time of year.
I seriously doubt you'll need to do that.  Regardless, if you are able to maintain a lossless collection of music, I think you should keep it.  Not simply because it is guaranteed to be artifact-free, but also because it will be easy to create a lossy library to whatever format and bitrate you like without having to transcode.

$2k amplifier
Perhaps even a $200 amplifier, depending on your power requirements.

Have you ever regretted ABXing?

Reply #28
I'm still de-programming as i think of it, and the though of taking my lossless file collection and turning them into a lossy format kinda makes me want to run for the hills.


Don't do that, BearcatSandor. Keep your lossless files (as storage, especially for lossless music, is very reasonable nowdays.) I'm sure you already know this, but using those lossy files as "source" to re-encode to the next great lossy would not be wise at all. If you can't hear the difference now you'll likely do so if you try that even once!

That's the benefit of lossless archiving: you can always go to the source, when the want arises, to get a transparent copy for your current needs.

You've actually inspired me to run some ABX testing myself. I'm finding I can tell the difference (most of the time) between a Lame-encoded mp3@128Kbps (CBR) and the FLAC...but I'm surprised by how negligible it is to me with my current equipment. I'm actually afraid to try 256 or 320 CBRs. I may shatter what remains of my audiophool ego.  (To be fair I do have a huge earwax problem that I need to remedy one of these days and am thirty-six years old. Fire guns on a regular basis and went to way too many heavy metal concerts in my youth.)

BTW "current equipment" is less than "audiophile" ideal. Alienware m15x laptop/Sennheiser HD 595s/Creative X-Fi Soundblaster external with the stupid fake 5.1 shit turned off. I am 100% certain that if I buy those $1400 "Hand-assembled in Germany instead of China" (whoopty-doo) cans that Stereophile is always raving about and a Grado headphone amp I'll DEFINITELY hear the difference!

If I had ten thousand US dollars to spend, I couldn't even get a recently-reviewed set of speakers featured in that fish-wrapper of a magazine.

...but yeah: I am finding that it actually irritates me that I don't hear a big difference even at the low bitrate that I've thus far ABXed with Foobar. Great post and thanks for the kick in the butt.

Edit: sorry, greynol. If I had been paying attention I'd have seen you covered that part about lossless archives being advisable above.
The Loudness War is over. Now it's a hopeless occupation.

 

Have you ever regretted ABXing?

Reply #29
No worries.

I think the fear that transcoding is certain to result in audible degradation is overblown, but nowhere near as overblown as the notion that audiophile-grade equipment will more easily reveal artifacts from lossy encoding.

Have you ever regretted ABXing?

Reply #30
I'm sure you already know this, but using those lossy files as "source" to re-encode to the next great lossy would not be wise at all. If you can't hear the difference now you'll likely do so if you try that even once!

Are you sure that it is "likely" to hear degradation from going lossless->"sensible lossy encoding"->"another sensible lossy encoding"? Depends on what is "sensible", I guess, but I would not dare to say that its likely. I'd rather say that chances of hearing flaws increase as the number of transcodes increase.

I also agree on keeping the lossless archive. Hard-drives are cheap, most of us dont have more than 1000 albums but ripping those 1000 albums a second time is no fun. It may not matter, but even the slight chance that it might matter is enough for me to keep the flac files. I guess that makes me an audiophile?

-k

Have you ever regretted ABXing?

Reply #31
I've noticed that people typically listen at much higher levels in the studio than they do at home
...and many of those professionals go very deaf very quickly!

Cheers,
David.


Have you ever regretted ABXing?

Reply #32
I wonder whether lossy will formats distributed over the net will soon become the only choice for mainstream music. Most consumers aren't that fussy and it'd make sense for the music industry. Personally, I've (nearly) only ever heard artifacts that were present on the CD. The couple of genuine artifacts were not in the least 'annoying' anyway.

So, I don't bother with ABX'ing anymore. I use LAME with V2 (or APS on old encodings) and make no attempt to find differences. I think that road leads to madness! I do have my CD collection archived to lossless though. Just in case...

Edit: I'm a Linux user too. You can ABX with Foobar, should you wish.

Have you ever regretted ABXing?

Reply #33
The only damage I've caused by abx testing has been to my ego.  No golden ears here, and no regrets (ok, initially some bruised pride).  I last did an abx maybe six months ago and it confirmed that I will probably never be able to hear any difference between lossless and either lame or oggenc at default settings,  and even if there is a difference to be heard I am extremely unlikely to notice it.  These days I notice differences between playback equipment but not between the sources.  That is an amazing difference from when I first heard mp3 about a decade ago.  If I hadn't occasionally done some abx tests I might have retained opinions based on unfavourable early impressions and experiences.

Have you ever regretted ABXing?

Reply #34
No worries.

I think the fear that transcoding is certain to result in audible degradation is overblown, but nowhere near as overblown as the notion that audiophile-grade equipment will more easily reveal artifacts from lossy encoding.


Has anyone actually ABX tested that premise? Just curious. I remember, many years ago, reading about a blind shootout between digital and analog in which the listeners were able to identify the difference on stacked Quads, but not on AR-LST's.


Have you ever regretted ABXing?

Reply #36
Surprisingly, to me, anyway, Fielder found that the noise of a quiet home listening room was below the threshold of hearing, and that the noise in an average room wasn't far enough above it to mask noise in a recording:


There seems to be a critical missing paragraph in the Fielder paper. It might be titled "Noise in recording Environments".

The paper as presented usecontains an unstated assumption that appears to me to be something like: "The dynamic range requirement for musical playback is irrelevant to noise in the space where the music is recorded."

IME, the space in which the recording is made is actually the weakest link. Large spaces are very expensive to make really quiet, and then you go and spoil the whole thing by putting performers into it. If you add an audience, then its close micing or lotsa noise, mostly both.

I see the Fielder paper as being a justification for HDCD, which we now know failed in the marketplace. IME the reason why is that due to the relatively high levels of noise in spaces used for recording, HDCD like SACD and DVD-A is a solution looking for a problem. It appears that SACD and DVD-A have also failed or are in the later stages of failing in the marketplace.

Have you ever regretted ABXing?

Reply #37
Has anyone actually ABX tested that premise? Just curious. I remember, many years ago, reading about a blind shootout between digital and analog in which the listeners were able to identify the difference on stacked Quads, but not on AR-LST's.

Not on this forum, at least.  It's a challenge I issue frequently to which no one has yet to rise.  Would you'd like to take a shot at it?

Anyone else?

Have you ever regretted ABXing?

Reply #38
Has anyone actually ABX tested that premise? Just curious. I remember, many years ago, reading about a blind shootout between digital and analog in which the listeners were able to identify the difference on stacked Quads, but not on AR-LST's.

Not on this forum, at least.  It's a challenge I issue frequently to which no one has yet to rise.  Would you'd like to take a shot at it?

Anyone else?


You know, I actually tried that many years ago, in a sighted bypass test using a PCM-F1 sourced off analog disk. To my surprise, I couldn't hear any difference. So I doubt I'd hear a difference with today's more advanced converters. Which of course doesn't mean that someone with better ears, electrostatic speakers, or headphones wouldn't (I did my experiment with Tympani 1-D's). But I did read recently someone else who had tried the same experiment with the F1, and was also unable to hear a difference. The acid test would I think be a comparison with a live mic feed, but since I'm retired from engineering I no longer have the facilities to do that.

Have you ever regretted ABXing?

Reply #39
There seems to be a critical missing paragraph in the Fielder paper. It might be titled "Noise in recording Environments".

The paper as presented usecontains an unstated assumption that appears to me to be something like: "The dynamic range requirement for musical playback is irrelevant to noise in the space where the music is recorded."

IME, the space in which the recording is made is actually the weakest link. Large spaces are very expensive to make really quiet, and then you go and spoil the whole thing by putting performers into it. If you add an audience, then its close micing or lotsa noise, mostly both.

I see the Fielder paper as being a justification for HDCD, which we now know failed in the marketplace. IME the reason why is that due to the relatively high levels of noise in spaces used for recording, HDCD like SACD and DVD-A is a solution looking for a problem. It appears that SACD and DVD-A have also failed or are in the later stages of failing in the marketplace.


Look again, it is there: Section 4.1, Noise in the Recording Environment.

Have you ever regretted ABXing?

Reply #40
I'm still de-programming as i think of it, and the though of taking my lossless file collection and turning them into a lossy format kinda makes me want to run for the hills
I don't think anyone would advocate converting to lossy simply because one can't hear the difference. A lossless archive has value as a future-proof 'storehouse' from which you can convert to other/better lossy formats at a later date. And of course, as you said, what if one later finds artifacts, and is stuck with them?

Have you ever regretted ABXing?

Reply #41
After a humbling fail on ABX with MP3 at 192kbp/s I reflected on how much "damage" was technically being done to the audio signal by the MP3 codec and I couldn't hear it (low pass filter, removing harmonics, higher noise floor, time domain anomalies, etc.).

With that as a context, the notion that audio cable, power cords, minute frequency response variations in amplifiers, gold CDs, etc. had much chance of being significantly audible seems really hard to believe.

I still look for competent measurements in what I buy, but I don't worry about it after that.

Room acoustics and speaker nonlinearities are typically orders of magnitude more audible than the others.  I believe they should really be the focus of the sound quality pursuit.  But they are complex, hard to reproduce and solve, so I suppose people gravitate to minutia like 16 vs. 24 bit recordings, etc.

By the way, I also vouch for keeping the lossless files so you can create versions in other formats, bitrates, etc. later.  I ripped my whole collection to MP3 the first time and then re-ripped to FLAC and use that as my source.  I highly recommend you keep your lossless source - ripping is so time consuming I believe you want to only do it once.
Was that a 1 or a 0?

Have you ever regretted ABXing?

Reply #42
I'm sure you already know this, but using those lossy files as "source" to re-encode to the next great lossy would not be wise at all. If you can't hear the difference now you'll likely do so if you try that even once!
Are you sure that it is "likely" to hear degradation from going lossless->"sensible lossy encoding"->"another sensible lossy encoding"? Depends on what is "sensible", I guess, but I would not dare to say that its likely. I'd rather say that chances of hearing flaws increase as the number of transcodes increase.

OK, I can happily agree with that. My wording isn't always perfect and I tend to sound "absolute" in print when I didn't mean it that way at the keyboard

I've never ABXed lossy-to-lossy trascodes and am going by things I heard on other people's stuff. (The type of people who just looked at me blankly when I tried explaining to them that taking an MP3 @128 and converting it to 320Kbps was not going to "improve its sound quality.)

Be that as it may, I've ABXed a few passages from certain songs over and over 100% positively @128Kbps LAME-encoded MP3. My point is I don't think it could get "better" if it was transcoded. The results are here:

Code: [Select]
foo_abx 1.3.4 report
foobar2000 v1.1
2010/11/05 00:36:34

File A: C:\dBpoweramp\EAC-FLAC\Jars of Clay\02. Unforgetful You.flac
File B: C:\Users\Engelsstaub\Desktop\New folder\02. Unforgetful You.mp3

00:36:34 : Test started.
00:37:46 : 01/01  50.0%
00:38:52 : 02/02  25.0%
00:39:34 : 03/03  12.5%
00:40:41 : 04/04  6.3%
02:07:57 : 05/05  3.1%
02:09:05 : 06/06  1.6%
02:19:47 : 07/07  0.8%
02:20:04 : Test finished.

 ----------
Total: 7/7 (0.8%)

(...I grew bored with this one after the seventh round. The passage was annoying to listen to over and over.)

I'm not taking some great pride in this because I do go out of my way to find passages that I "just know" I'll be able to ABX and, as before, I'm humiliated at how subtle the difference is. I have to concentrate very hard to hear such "differences" and wouldn't likely notice a thing if I were just listening to them without prejudice. That bugs the crap out of me...I realized that I could actually live with it if I had to.

I used to believe (until recently) that "even at higher bit-rates the high-hats wash out!" But I have heard this effect on transcodes. Truth is, I never heard "the high-hat thing" on my MiniDiscs in the nineties. I knew nothing about ATRAC compression then and believed that these were basically 1:1 digital copies via fiber optic transfer. ...but I know there were a (very) few instances where I was like "what happened to this effect?" or whatever.

...so anyway: I was just thinking "logically" that what was nearly transparent before would likely become apparent after. My bad for making assumptions and claims without more than anecdotal proof. Sincerely.
The Loudness War is over. Now it's a hopeless occupation.

Have you ever regretted ABXing?

Reply #43
I've never ABXed lossy-to-lossy trascodes and am going by things I heard on other people's stuff. (The type of people who just looked at me blankly when I tried explaining to them that taking an MP3 @128 and converting it to 320Kbps was not going to "improve its sound quality.)

Be that as it may, I've ABXed a few passages from certain songs over and over 100% positively @128Kbps LAME-encoded MP3. My point is I don't think it could get "better" if it was transcoded.
What you think is irrelevant. Simple mathematics dictate that lossy-lossy transcoding cannot improve sound quality, in the context of perceived fidelity to source; that is, each successive transcode creates an audio stream further removed from the original. Unless one idiosyncratically prefers ringing/washing artifacts, etc., I imagine that lossy-lossy transcoding cannot improve sound quality by any definition of the term.

Have you ever regretted ABXing?

Reply #44
What you think is irrelevant. Simple mathematics dictate that lossy-lossy transcoding cannot improve sound quality, in the context of perceived fidelity to source;


There is no mathematics of perceived fidelity. All there is is a pile of empirical findings over a small subset of the population which has lead to the couple of very successful (in regard to a much larger subset of the population) implementations of lossy encoders, we have today. This raises the hope that inductance was justified, but we are still very far away from the features usually attributed to mathematical assertions.

Actually it is indeed possible that a lossy transformation improves perceived quality (whilst certainly not the rule).

Have you ever regretted ABXing?

Reply #45
Yeah, I may have overgeneralised.  I think I meant to refer to absolute fidelity, not perceived. Thanks for the input!

Have you ever regretted ABXing?

Reply #46
Which of course doesn't mean that someone with better ears, electrostatic speakers, or headphones wouldn't (I did my experiment with Tympani 1-D's).

Ignoring the part about better ears (higher sensitivity to lossy artifacts), this doesn't mean that there can't possibly be a pink elephant orbiting Uranus, either.

After what I've read from people who have actually developed lossy codecs and those that understand lossy encoding on a level far greater than I, it is my understanding that all other things being equal, the farther the deviation from a flat frequency response, the more likely one is to hear lossy artifacts.  I've also read that electrostatic speakers don't provide a very flat frequency response, so assuming that someone can more easily distinguish a lossy encoding with these speakers (and again I've not seen any objective tests that demonstrate this), a wildly uneven frequency response would be the first explanation that would come to mind.  Same goes for "better" headphones.

Have you ever regretted ABXing?

Reply #47
Which of course doesn't mean that someone with better ears, electrostatic speakers, or headphones wouldn't (I did my experiment with Tympani 1-D's).

Ignoring the part about better ears (higher sensitivity to lossy artifacts), this doesn't mean that there can't possibly be a pink elephant orbiting Uranus, either.

After what I've read from people who have actually developed lossy codecs and those that understand lossy encoding on a level far greater than I, it is my understanding that all other things being equal, the farther the deviation from a flat frequency response, the more likely one is to hear lossy artifacts.  I've also read that electrostatic speakers don't provide a very flat frequency response, so assuming that someone can more easily distinguish a lossy encoding with these speakers (and again I've not seen any objective tests that demonstrate this), a wildly uneven frequency response would be the first explanation that would come to mind.  Same goes for "better" headphones.


That's interesting. However, while I've seen response curves of electrostatics that were quite messy, the original Quad, which was used in the A/B test to which I referred earlier, had a pretty good on-axis response:

http://www.nutshellhifi.com/MLS/MLS2.html

Not bad for 1957! (The bass rolloff is, at least according to the web page, a measurement artifact.) At the same time, it was unusually directional in the high frequencies, and I don't know how regular the polar response was.

Here are AR's curves for the LST:

http://www.classicspeakerpages.net/library...ochure_pg2.html

Note that they intentionally present only the individual driver response rather than on axis curves. The on-axis response of the AR's of that era was significantly irregular at the crossover points as a consequence of primitive driver placement and crossover designs.

Electrostatics tend to have very low measured distortion and unusually good impulse response. I don't know whether and to what extent that may have contributed to the test results. I'm also not aware of any experiments that might confirm or refute their reputation for being unusually revealing of distortion or artifacts in source material, or identify factors that might contribute to that reputation. It sounds like an interesting area for research.

Have you ever regretted ABXing?

Reply #48
There seems to be a critical missing paragraph in the Fielder paper. It might be titled "Noise in recording Environments".

The paper as presented usecontains an unstated assumption that appears to me to be something like: "The dynamic range requirement for musical playback is irrelevant to noise in the space where the music is recorded."

IME, the space in which the recording is made is actually the weakest link. Large spaces are very expensive to make really quiet, and then you go and spoil the whole thing by putting performers into it. If you add an audience, then its close micing or lotsa noise, mostly both.

I see the Fielder paper as being a justification for HDCD, which we now know failed in the marketplace. IME the reason why is that due to the relatively high levels of noise in spaces used for recording, HDCD like SACD and DVD-A is a solution looking for a problem. It appears that SACD and DVD-A have also failed or are in the later stages of failing in the marketplace.


Look again, it is there: Section 4.1, Noise in the Recording Environment.


No, data about totally empty rooms is *not* relevant, at least until we start generally making recordings in totally empty rooms.  I think there's only one piece of music that could be recorded this way. It was written by John Cage, if memory serves.  ;-)

There are also quieter rooms than those. I think one of them is in the Carlsbad Caverns in New Mexico. If data is going to be relevant it has to have a useful level of generality.

Have you ever regretted ABXing?

Reply #49
Electrostatics tend to have very low measured distortion and unusually good impulse response. I don't know whether and to what extent that may have contributed to the test results. I'm also not aware of any experiments that might confirm or refute their reputation for being unusually revealing of distortion or artifacts in source material, or identify factors that might contribute to that reputation. It sounds like an interesting area for research.


There have been enough mediocre and even  *bad* electrostatic speakers pushed onto the market that trying to characterize electrostatics as a class of Spears as having any unique sonic advantages is very questionable. Even the Quad ESL curves you reference have a rather obvious flaw - lack of what most of us would call bass response. Some of that might be due to the measurement environment. I just can't tell from the accompanying text.  The Quad electro stats of that era also have some pretty strong dynamic range limitations.

It is no accident that the most accurate of currently available loudspeaker systems are usually direct radiators with cone and or dome drivers.  Furthermore accuracy is always strongly limited by the room, and the matching of the speaker system to the room.  So raw speaker system measurements are not sufficiently representative,