Hiya,

I spent a good chunk of last night... well the whole night absorbing information from this site. I stumbled here from the ever controversial topic of mp3 < CD.

I finally am making myself familiar with ABX software and ran a few tests without headphones and it was very hard for me to tell the difference beyond 128 ripped with lame.

To keep the hardware and variables as close as possilbe here is the equipment i am using.

Computer CD Transport.
Audigy 2 ZS MP3+
Coax digital out fed to...
Arcam 300.

So far i have done only speaker only test i will spend a few hours doing some tests with some sennheiser soon but my immidiate question is.

- Do most of you do abx tests with ABX only from soundcards analogue out? And do you think or suggest there is a big difference in my reciever doing the decoding?

I had also found previously that digital out quality differed from one model to another as my previous Audigy was not upto par with my current Audigy 2 ZS. I am contemplating going in market for a maudio to get optimum out of my digital out.

Thanx

That's a hard question to answer. A lot of good ABX results have been obtained from hardware that audiophiles may sneer at. However, when there are known and probably-audible problems with the hardware, the ABX results could be skewed towards either reduced audibility or increased audibility. It entirely depends on the nature of the hardware. Examples abound of both situations.

FWIW, when I have done ABX tests, I have generally done them from either the analog out of my sound card, or through a headphone amp. In my experience, the blind-tested differences between sound cards, when they are using reasonable oversampling techniques and output impedances, are such that it's very hard to ever conclusively tell whether one analog out really is audibly worse than another. So the differences are so small that the differences present in the ABX samples probably dwarf whatever distortions exist in the sound card.

That's not discounting the more obvious and stupid problems. The resampling present in some Creative Labs cards is such a problem, and it's still somewhat controversial. (IIRC, a couple people have piped up with successful ABX tests against the AC97 resampling code.)

The usual fix is to buy a $20 Chaintech AV710, or any other dirt-cheap Envy24-based sound card with digital out, or if you're particularly spendy, an Emu 0404 or 1212m. Of course, one may complain about their jitter specs, but like with the differences present in sound card analog outs, I've never seen anybody convincingly state that such differences are audible. Few people who complain are competent enough to even measure such things.

Thanks for the input.

My goal is to find optimum bitrate for cds -> mp3 conversion so i dont have to keep on swapping discs to listen to various songs...

My side goal is also to get the most optimum sound out of the pc to the reciever.

So i set up some files which were encoded as follows:

Encoder: Lame 3.92
Mode: Stereo
Quality: Highest Quality
Sample: 44100 Hz
Bitrate: VBR 0 (highest)
Min: 32kbit/s
Max: Variable
VBR Mode: New VBR
Lowpass: Disabled
Highpass: Disabled

I am trying to get my hands on a sennheiser but by listening to the song with precussion only i can clearly tell a difference at 128... 160 and above is hit n miss which is mostly guessing.

I think to get everything alligned i for sure need to use headphones and connect it to my arcams headphone jack to compare...

IS that how you guys do it or what?

p.s my speakers are custom made they are kinda like the totem speakers with upgraded caps n tweeters n such.'

I was finally able to get my hands on a entry level sennheisser HD 555.

I was able to tell difference between 160 kb and above that was guessing.

So do you guys reckon that:

- if i used the listening samples instead of actual cd songs i may have better results or
- my experiment is in line with other peoples findings or
- is my headphones the weakest link?

even if all above is true and its hard to differntiate on listening to music for music does it really matter that the differences that are < 5% be so much that it matters to spring out for a cd changer or rip all to htpc.

thanx

If you want the optimum bitrate, use the latest version of LAME, and use the parameter -V 2. This used to be the setting --preset standard. Further reading is available on the HA Wiki page

You may have better luck with the samples. IIRC, in they're sometimes used as "tough to encode" samples (though not always).

If it makes you feel better, rjamorim's public listening tests were conducted at lower bitrates because at higher bitrates, it becomes more difficult for the average person to hear differences (amongst other reasons). I personally have a challenging time with listening tests because too often I find myself listening to the music itself rather than listening for the differences.

I can't say whether your headphones are the weakest link.

As for whether or not it matters, that's a question that only you can answer. For some, it does matter, to the point where they encode all their music into a lossless format. Others go with an alternative format in order to save space. You can certainly solicit other people's opinions, but all you will get is opinions.

It is very hard to hear differences between good lossy and lossless formats. You're on the right path by using coax digital out.

I will say that I have heard a clear difference between digital outputs feeding the same DAC. Comparing a Firewire MOTU 896HD via AES/EBU to a USB Emagic EMI 2|6 interface via S/P-DIF coax, I can clearly hear a better sound from the MOTU. I am not able to remove all variables, of course. There could be differences between the AES/EBU cable and the S/P-DIF cable, although both are of similar length and are probably made by the same company. I suspect that the real difference is that Firewire is superior to USB, but I can't really rule out that the digital switching path between the AES/EBU and S/P-DIF inputs on my DAC could be dissimilar.

All of this is my way of showing that problems with your coax digital output (most likely jitter) could be adding distortion to the sound which masks any problems with the lossy coding. Psycho-acoustic studies have shown that humans often prefer (slightly) distorted audio for the same reason they prefer louder audio - and this is primarily related to the fact that distortion adds harmonics and thus adds decibels to the perceived level.

Personally, I would recommend that you focus on good speakers, careful placement, time alignment, and also a good Firewire audio interface with coax digital output (or AES/EBU if your Arcam 300 has an input). In my experience, MOTU know what they're doing. Without a good listening system for your ABX tests, your best bet is to just use lossless formats or maximum bit rates.

Digital data is digital data. Except if your usb card manipulates the sound, it should be identical to the decoded audio stream. The data should be transmitted without problem over either USB or Firewire, both having more than the required bandwidth for standard audio streams.

thanx for the input.... i think i have narrowed it down to if a digital file is decoded properly it can be very clear and in comparison to cd pretty close... given if you were keenly listening for differences and even then it would be hard for ME atleast...

For me a good 192 wiht latest lame full stereo encoding is very hard for me to pick the mp3 i clearly fail at those tests.

The whole purpose was this for me was to the one directional comments from lots of retailers, me being from the audio video industry i wanted to explore the mp3 domain.

I think at a personal level it is probably very arrogant and ignorant of people to blindly call mp3 "crap". Given mp3 is by design an inferior format, but i have done A/B tests even with those said audiophiles who shrug at the thought of mp3s and sadly they failed to differntiate at around 160 and 192 around the same level i did.

Offcourse there will be some keen listeners who with thier headphones can pin point a 192 9 out of 10 times but i think for the sole purpose of audio quality thats not deterioted by compression mp3 probably is a good compromise.

any thoughts?

p.s my audigy is a PCI card.

You might want to look at some listening tests: http://wiki.hydrogenaudio.org/index.php?ti...Listening_Tests

These are mostly quite old, and by intention very demanding, so they should reassure you that your inability to pick mp3 at ~160-192kbps is absolutely normal.

Me, I can't ABX LAME at 128, but I'm old.

Best

Michael

If you're asking "what should I encode at so that the mp3 sounds the same as the CD almost all of the time" then read the wiki or FAQ...

http://www.hydrogenaudio.org/forums/index....showtopic=28124

...it's there for a reason. Making up your own settings is not the way to go. Unless you know the encoder inside and out, you'll probably make things worse even if you think you're making them better. Your attempts at disabling filtering and forcing discrete stereo etc are classic examples of this.

The vast majority of very picky people are very happy with the recommended settings. There are some individuals with some types of music which are not "mp3 friendly". Every audiophile thinks they must fall into this category. I've yet to meet one that actually does, but several HA members do.




Another thing: the analogue output of some sound cards is rather noisy. I've also seem people report the analogue output of very good sound cards being noisy in certain configurations/systems.

For CD quality listening, a digital output is a great idea, but beware that the combinations of some computer programs, some low quality audio (e.g. 11.25kHz sampled), some windows output methods, most sound cards, and some outboard DACs will give you no audio at all, or really horrible audio.

Cheers,
David.

Thanks to you all for some great pointers and comments.

i will check out the links and try the pre-designed samples to do abx.

I found true stereo is better than joint stereo... lot of encoders default to joint stereo i am not 100% if lame does that but for safety i specify stereo and go from there.

p.s also will try different soundcard outs (digital only) to see if that makes a difference....