I think that its important to note that we are not all on a level playing field as far as ABXing goes. I, for one, am using a Turtle Beach: Santa Cruz sound card:

http://www.turtlebeach.com/site/products/s...producthome.asp

Hooked directly into a NAD C350 Intregrated Amp with a super high gage Y-adapter cable:

http://www.nadelectronics.com/hifi_amplifi...350_framset.htm

With Mirage FRX-3 Main Speakers and a Mirage FRX-S8 Subwoofer:

http://www.miragespeakers.com/products/cur...frx_3/index.htm
http://www.miragespeakers.com/products/pas...ducts/index.htm

This gives me a pretty good edge on ABXing, I don't think I could tell a 192 LAME CBR from the origonal wave if I was using a dinky pair of computer speakers. As it is I can tell LAME from WAV most of the time even when using the command line options: -V 0 -b 128 (gives you VBRs usually between 260 - 305kb/s).

I use MusePack on --quality 8, even though I can't currently ABX any samples I have tried on --quality 7 just because I feel sometime down the road I might be kicking myself for not using a high enough quality. (like I am now from my LAME -q 4 days)

Soundcard : Creative Labs Audigy
Amplifier : Yamaha 592
Headphone : Sennheiser HD570
Fileformat : FLAC

Comments :

I'm probably going to upgrade to a better soundcard and I'm going to build me some speakers. I'm still choosing between the diy proac response 2.5 clone or the diy tempo clone. I've recently changed from mpc as my main choice for archival to flac, simply put : it's lossless so no more listening for artifacts and such. I'm still using mpc though when I've got some music to take to friends.

Links :

diy proac response 2.5 clone
diy tempo clone
Yamaha 592
Sennheiser HD570

Soundcard: Stereo Link USB
Acoustic Research TDS-202 Sound Enhancement (info)
Amp : AudioSource Amp Two
Speakers: Infinity Alpha 40
Fileformat: MPC --quality 8 --ms 15 (because I think lossless is too inconvenient to store for 1000 cd archive)

All cables from Knu Konceptz.

About ABXing, the importance of equipment for hearing artifacts is already stated to be dubitable on these forums. And headphones would probably serve much better (if the problem is not associated to soundfield that requires an ambience)...

I've done majority on my ABXing on SB Live Value and some old studio headphones (modified).
These cans don't keep their quality standards anymore...
Additional equipment include old HiFi integrated and old HiFi loudspeakers,
which start to fall apart. (both of Polish origin, high-end models '85)

You see, my equipment doesn't set new records in quality,
but I can still ABX most of the samples using it.

Recently bought new Terratec Aureon 7.1 and Sennheiser HD497.
The phones should arrive later this week.

Anybody knows a method to do a double-blind test with headphones?

Lossless isn't an option for me, as I have nearly 9000 music files (which currently take up over 50GB on my 120GB drive. As far as your particular choice in lossless audio, I have heard that flac is pretty old and totally inferior to monkey's audio.

Studio headphones are still probably pretty good even if they are old, and as for the SB Live . . . well at least its not integrated .

I'd be very careful to state such things in HA. Because flac is neither old nor totally inferior to Monkey's Audio. It has been actually discussed very recently. I favor and use Monkey's because of APL support (although buggy), but am very upset with the lack of replaygain support which is essential for my Waves DSP plugins to work.

Did I mention Adapt-X combined with Waves DSP plugins (MaxxBass, REQ, L2) improved my pleasure out of my system double fold? So I would say it's not only hardware that matters.

Time to dispell some FUD...

1. FLAC isn't old, check their webpage.
2. Monkey's Audio really compresses better but much slower and requires lots of cpu power to decode. Additionally it doesn't have any hardware support and won't have in the near future.

OTOH...
I've listened to Senn PX200 and they seemed slightly more neutral and detailed than my old phones.

man oh man do some people have lot's of music, how do you listen to it all? I've only got a couple of albums encoded in FLAC atm so theres no big problem for me if I decided to change. Where did you hear that FLAC is inferior to monkey's audio? I know that monkey's audio compresses a bit better in general but I just don't feel good about it, for example I've heard that there have been a couple of changes in the fileformat breaking compatibility. I've read something about it going opensource but does this mean it's totally shifting towards something like FLAC, with that I mean : fully open, stable fileformat, crossplatform, ...
Maybe if someone could give some answers on these issues I'd reconsider, but for now I for example have no idea what it would mean to monkeys audio if Matt (the leaddeveloper) would stop/kill the project.

For ABXing, NOT listening, I use a laptop and a pair of Aiwas. I've gathered this much, this is my own opinion from experience:

1) The soundcard makes little difference, compared to the headphones. Any soundcard will do really, the big difference comes in integrated ones, mainly laptop ones, since there is a lot of background hissing added, so as long as it doesn't add large amounts of unwanted noise to the output, it's good.

2) It matters what league the headphones are in, not what headphones they are. Example; expensive sennheisers(I think it was the HD580) and those $40 Aiwas above made no difference in ABXing, but the $20-$30 behind your head lightweight sonys, made it impossible for me to ABX a 128k MP3 of my normal music. You can tell just with a quick listen if the headphones are in the HQ or LQ league, you don't need super headphones to ABX something, just something that will play the music without distorting it like ultra cheap ones that come with your CD player as freebies do.

And unless it's a bad killer clip, speakers are nothing for ABXing. Hook all that up to a pair of decent headphones, then try, although MPC is pretty unabxable in normal music at -q5 to begin with. Whatever you're ABXing now will be way easier with headphones.

I completely agree with Diff on this one. I've been doing a heap of ABXing of late with Wavpack lossy, Optifrog Dualstream and a bunch of other formats. My gear for this is more often than not, my IBM Thinkpad's CrystalFusion on board sound, through a pair of cheap Sony street style 'phones that came with my Sony MD.

The key to ABXing is not the fidelity of the recording, but being able to hear actual differences in the music, a smeared attack here, a strange squeak there.

At other times, I use a pair of other Sony head phones whose number escapes me at the moment, through the SB Live in my desktop PC I assembled at home.

I generally get the best results from the laptop though, as it is quieter than my desktop PC, and I can take it somewhere quiet and concentrate on the testing without distractions. B)

The level playing field referred to above is meaningless unless the listener knows what to listen for. If you really know what artifacts sound like it's amazing what you can sometimes find. I have before burnt a heap of decoded wavs onto a CD, and been able to ABX a bunch of various formats on the way to work with my car CD player, which is not exactly what I would call high end audio.

I would be surprised if there is anyone who couldn't ABX fatboy with LAME 3.90.3 @ 320kbit, even in a car stereo, for example.

Den.

Thanks for the tip on ABXing, its too bad my good headphones recently died. I do know that there isn't THAT much difference as long as you are using a PCI sound card and some other at least decent gear, I just thought that there might still be some difference . . . . . . although it makes sense to me that there wouldn't be none that I think of it.

Anyway while we are on the topic of ABXing I thought I would bring up another beef of mine about the way some people ABX around here. I have seen a few people, for example, pick a --quality to set MPC and then try to match the exact bitrate of the output file with an Ogg. This doesn't make sense to me because of the nature of VBR, one codec will ALWAYS find a particular track easier or harder to encoded then the other. Thus one of the encoders is being short changed by having the bitrates of each match exactly. Wouldn't it be more fair (if, at the same time entirely less practical) to try to encode many, many files at a certain --quality then encode the same many, many files in Ogg and fine tune the -q until the average average (not a typo) bitrate of all the MPC songs were the same as the average average bitrate of all the Ogg songs?
I imagine I lost a few of you here with my wording. If anyone actually understands this and feels that they can explain it in clearer english please, feel free.

Either Wavpack with those clips it was completely mangled, or you're a bat. The only thing I can say for those headphones is that they destroy the sound, they sound as if someone took the good headphones and had a two year old scramble the EQ around randomly.

It takes time to adapt to different codecs. I remember when a 128k MP3 sounded transparent, then I took out a hard instrumental guitar/drums rock song(easiest to ABX to me) and it sounded bland and wrecked the high frequencies which I never heard before, then I tried Ogg Vorbis at -q0 and it sounded transparent, then after awhile even -q4 wasn't good enough for that same song. That's a reason I don't like "training" sites, since they ask me to look for what they hear, not what I hear.

At that time everyone I talked to seemed to think that 128 CBR was lossless, I was one of the only people I know who was saying "CD quality???? I don't think so . . . " the first time I ever heard an mp3. 128 has always sounded bad to me.

I guess I really don't know anything about ABXing. I have no idea how to listen for specific artifacts, I just go by what feels right and what feels wrong. With this, low class technique I do believe that the equipment you use does make a big difference. Ok . . . . . so in light of this, what does pre-echo sound like?

I agree with you. When I first started posting my test results here, I was slapped by the regulars, claiming that unless the bitrate is the same, it's like comparing apples and oranges. I now post tests at the same bit rate to keep people happy, but I actually prefer to compare formats using the quality setting, or bitrate commonly used out there for a particular reason, ie portables, archiving, etc.

If you have the same bitrate though it does stop people from using bitrate as an excuse. You can imagine that if you posted a test where a Vorbis file at 160 kbits, was soundly beaten in quality by an mp3 at 180 kbits, the Vorbis fans would blame the lower bitrate and rule the test invalid. If you go for similar bitrates, it prevents people using such an excuse... B)



Absolutely, in terms of the crappy head phones. Wavpack is not the best example as it usually doesn't cause artifacts, it mainly just adds hiss. While the phones I sometimes use are crappy, they are consistently crappy if you get my drift, so I can still readily pick differences when I do a back to back comaprison of an encoded file to the original wave file, assuming they've been replaygained appropriately, etc. In Wavpack's case, the hiss changes in it's pitch, and I find it easy to pick. Both David (Wavpack) and Florin (Optifrog) are busy at it though so things are only going to get better!

I also have never visited a training site. Just over time, with repeated listenings, you start to notice things as you say, and before you know it, it becomes quite easy. The things I hear don't always seem to quite match what others describe they are hearing, but I can ABX it nonetheless. It was funny reading the comments from the AAC 128 kbit test recently. For some samples, I think every participant described hearing something different, so it made you wonder if there was actually anything of the original wav file still recognizable in the encoded file!

Den.

To listen for artifacts, "train" yourself.

Listen to an MP3 of a complicated song, at around --alt preset CBR 80, listen to what makes it sound bad, then go to 112k, then 128k, and up from there. Do this with a few different complicated songs, then you'll know what to look for. 80k-112k are the places I'd recommend to "train" yourself with, with MP3, since the artifacts are there, but a lot more subtle than with below 70k so you'll know what exactly it's screwing up with.

Pre-echo... listen to something with high hats and lots of highly audible cymbals at 80k or something low. The way it screws up, sounds like pre-echo.

Yep.

Listen at a low bitrate, get a feel for how the encoder messes up your chosen sample, then try at a higher bitrate. This works well for me also. B)

Unfortunately this is not so easy for some trying Wavpack/Dualstream as the encoders don't go below ~265 kbits. So if 265 kbits is transparent, you are screwed!

Thankfully this is not the case for me.

I don't know how many other people have this problem, but when I listen to an APE file played from the Phatbox in my car, I can't seem to hear anything, no matter how high I turn up the volume, whereas FLAC sounds just fine!

JUST KIDDING! APE is not a compatible format for 1/2 of my primary listening platforms (i.e., my car) while FLAC is. I'd venture to say that's the primary reason many people currently choose not just FLAC over APE, but also MP3 over MPC, etc. *Hardware Compatibility*, rather than anything to do with sound quality or compression/decompression rates. No matter how good it sounds on my PC (or fast is decodes), if I can't play the format elsewhere in my life, then I can't use it at all. If I had the storage capacity to keep multiple versions of the same music, then things might be different, but 20GB x 2 just isn't cutting it.



I can't, and I estimate that it's transparent in my car as well, which has a much better sound system than my PC. And to add my own PC hardware list since everyone is sharing...

-PC: Sony VAIO GRX670 (notebook)
-Soundcard: Yamaha AC-XG
-Amp: huh?
-Speakers: Klipsch THX 2.1 (w/ 170watt built-in amp)
-Cans: Cheapies...but the best portables I've owned...Sony MDA-ED228 earbuds, actually sufficient for *some* ABXing. Will soon be buying some nice Sennheisers or equivalents, though.
-Encoding Format: LAME --alt-preset insane (v3.90.3) for primary, --alt-preset cbr 128 for portables (transcoded from --api...frowned upon by many, but still close enough to transparent for me coming out of my so-so MP3 player and buds)

I think the argument that good equipment is important for ABXing is a valid one, but it depends on *what* you're trying to ABX I would think (speaking as a newbie here). I could ABX Kali without too much trouble on earbuds, but I'll bet those of you with much better hardware could find more discrete differences that I simply wouldn't be able to detect. I also agree that using headphones to ABX is better many times than using speakers. I was ABXing at 18/20 or better with the earbuds on some samples that I couldn't get even 12/20 on my beloved Klipsch's.

Also, den makes a good point about using a notebook PC to travel to an "ABX-friendly" environment. To me, that would seem better than using higher-line hardware in an environment with too much noise or distractions. I've used mine to escape my g/f when she's nearby on the phone while I was trying to ABX, for example.

Anyway...all I want out of life right now (other than world peace and goodwill towards all people) is a good set of headphones and a 120GB USB 2.0 external hard drive. Oh yeah...and one of those seperate amplifier thingys...

Sorry, maybe it was a different codec I heard about . . . . . Anyway I am done posting things that I don't know about for sure . . . .

No harm done! But indeed, those things tend to backfire on you here

Soundcard: original SoundBlaster 16 ISA <--- for you youngin's, that was before PCI

NAD 7140 Receiver
Dahlquist DQM-9 Speakers

time for a new soundcard? or will it really not make much of a difference...

Soundcards:

M-Audio Audiophile 24/96 (currently running to a pair of Denon AH-D550 headphones, soon to be Sennheiser HD280 Pro's... they should come in today)

Soundblaster PCI128 (ES1371) currently running to a set of Altec Lansing ACS45.1 speakers

External:

Thorens TD-160 MkI turntable (ca. ~1974) with Audio-Technica 440ML cartridge (to a Technics SU-C01 vintage preamp ca. ~1979, then to the inputs of the Audiophile 24/96 card)

Internal:

Plextor Plexwriter 24/10/40A CD-RW
Toshiba SD-M1612 DVD-ROM

That's about it... I may be picking up a headphone amp pretty soon too, but haven't decided whether to go "pro-audio" (studio type stuff) or "audiophile" (pricey type stuff ).

I enjoy "mixed" threads like this that are a combination of "showing off" (equipment) and getting across some useful information too... lots of "nerdy" type fun... As far as ABX'ing anything, I can hardly remember when was the last time I did it (altho I'm very appreciative of the hard work of others in improving MP3, as well as other codecs).

SB16? Where can I buy one for old games?
Damn, my mainboard has no ISA slots... (but other machines still have them )

The difference will be quite large, especially if you move to 24bit card.
(no, Audigy1 not included)

@fewtch: Build one yourself
Either Xin Super Mini or famous Meta42 or CMoy

Noooooo... help! Hot! Soldering iron hot!

Actually I kinda have a thing about using my own homebrew stuff (whatever it is)... somehow, I feel much better with the commercial stuff that I can't screw up. Unless it's something very basic like a switchbox of some kind (done a few of those for antennas when I was into shortwave radio).

Time for a new motherboard if your still using ISA . . .