Some Sound City Guidance

[DRZ]

I myself am no audiophile, but I have spoken time and time again with people like yourself; all say that compressed audio - in any form - is nothing like the origional source recording when played back on a decent Hi-Fi. A lot of information is lost in 128kbps MP3s, and there's no denying that - hence the term compression.

No, the term compression doesnt imply losing anything at all. When you zip (a form of compression) a word document, do you lose half of it? Nope.

There are two types of compression, Lossy and Lossless - Lossless is clearly superior to Lossy mathematically but if the Lossy method uses a good algorithm, it is rarely discernable by the human ear.

For what its worth, Lossy compression uses techniques to remove things that are deemed to be imperceptable - done well, this is the case, you just cant hear it.

A lot of "Audiophiles" are very quick to just jump on the "MP3s are rubbish" bandwagon and are quite suprised when they are unable to correctly tell which is which

Honestly, get someone into an anechoic chamber and start REALLY testing thier beliefs and 9 times out of 10 it all falls apart and they leave unhappy people.

 

[james.miller]

i believe it was DRZ who conducted a blind test on these forums, where something like 6 samples of a piece of music were post at various encoded rates, and we had to try and guess which was which. the results were nothign short of stunning. Virtually nobody got it right, and very few people could tell the difference between them.

It's an eye opener, and the reason i now am totally happy with encoding my own tracks using high bitrate VBR Lame and not resorting to lossless codecs.

 

[TooNice]

@james.miller: I can see how that test would be difficult.
- Are the users told which track is uncompressed? Familiarity with the track help in those tests.
- It is one thing to tell appart the difference between two track (something I would expect someone be able to do when the difference is between 128kps and uncompressed), and to tell how they are encoded.
- Of course, take LAME@APS vs LAME@API and I would not even think of telling those two appart.

 

[DRZ]

i believe it was DRZ who conducted a blind test on these forums, where something like 6 samples of a piece of music were post at various encoded rates, and we had to try and guess which was which. the results were nothign short of stunning. Virtually nobody got it right, and very few people could tell the difference between them.

It was me, yes - and IIRC, you were the ONLY person on this forum that was tested that got it consistently correct at low bit rates.

Many "audiophiles" with significantly superior equipment to yourself have failed many times to accurately tell me what was what.

Yet I still have people with PC speakers telling me it is abundantly clear which one is the low bit rate.

 

[DRZ]

@james.miller: I can see how that test would be difficult.
- Are the users told which track is uncompressed? Familiarity with the track help in those tests.
- It is one thing to tell appart the difference between two track (something I would expect someone be able to do when the difference is between 128kps and uncompressed), and to tell how they are encoded.

It was the same 30 second sample. Nobody but me knew which was which

I used 96k, 128k and 192k CBR, WMA lossless and some 220k+ VBR compression rates, MP3s encoded using the fraunhoffer algorithm IIRC.

Suprisingly, across the 3 tests I have done, more people picked the WMA as the original than any other of the tracks

 

[TooNice]

I used 96k, 128k and 192k CBR, WMA lossless and some 220k+ VBR compression rates, MP3s encoded using the fraunhoffer algorithm IIRC.

Suprisingly, across the 3 tests I have done, more people picked the WMA as the original than any other of the tracks

Wait, doesn't that suggest people can tell lossy and lossless appart? o.O Or am I misreading something (do you mean the VBR ones are also WMA?).

 

[DRZ]

Wait, doesn't that suggest people can tell lossy and lossless appart? o.O Or am I misreading something (do you mean the VBR ones are also WMA?).

Nope - the way blind tests work are that there is a requirement that the subject gets everything right for it to be considered reliable. If they cant tell the difference between a 220vbr and the original yet think the WMA is better than the both, its obvious that they didnt actually know at all or it was a guess, therefore they didnt know at all

I had to go through all of this when trying to find the JND and JNND for a test of the smallest perceivable difference in SPL. I ended up binning most of the results because they were total garbage.

It is amazing how people would rather say they can hear a difference rather than admit that they cant!

 

[james.miller]

It was me, yes - and IIRC, you were the ONLY person on this forum that was tested that got it consistently correct at low bit rates.

Many "audiophiles" with significantly superior equipment to yourself have failed many times to accurately tell me what was what.

Yet I still have people with PC speakers telling me it is abundantly clear which one is the low bit rate.

LOL. i didnt actually realise that.I'm not even using a proper amplifier lol. It IS true though, many people do claim to have an ear for music, backing that claim up with their high end equipment.

I've stunned everybody who has listened to my setup, not only because i'm 'only' using an av amp to power the main speakers, but also because im using a custom built 12" subwoofer (well, 2 of them) which always stays on with music. Apparently i do everything wrong but in the end it just works, while everybody with their high end equipement has left here in amazement.

The main thing to remember is dont believe what you read in the magazines or on forums. Blind testing will bring almost everybody down, and a proper lame encoded mp3 is hard to fault, whatever the equipiment you use

 

[Phil99]

For gaming:

Creative > *

For Music:

M-Audio/HDA etc > Creative

For audiophile usage you don't want to be buying a Creative card, but for gaming there really is a difference with Creative (EAX3, 4, 5 etc) vs EAX2 cards as many don't even have hardware audio acceleration.

That's speaking from personal experience; I've yet to find a card that can beat the Creative lineup in terms of gaming...managed to get a soundstorm board free this week and it's better than the C-Media 8768+ based cards (X-Mystique) with regards to gaming as it's hardware accelerated, but still not up to the Creative standards in games.

 

[DRZ]

I've stunned everybody who has listened to my setup, not only because i'm 'only' using an av amp to power the main speakers, but also because im using a custom built 12" subwoofer (well, 2 of them) which always stays on with music. Apparently i do everything wrong but in the end it just works, while everybody with their high end equipement has left here in amazement.

The main thing to remember is dont believe what you read in the magazines or on forums. Blind testing will bring almost everybody down, and a proper lame encoded mp3 is hard to fault, whatever the equipiment you use

Anything reasonable that is well set up delivers staggering results 7 times out of 10. Buying expensive kit is easy and letting the dent in your bank balance (or pride, should it turn out to be rubbish) convince you that it is amazing is even easier.

Setup is absolutely critical and from your general posting here I would hazard a guess that is all on the button.

Again, you are correct about blind testing. Putting someone in the true unknown and forcing them to make a decision really sorts the men from the boys. I found my first blind test in proper conditions to be quite distressing actually. Having your assumptions turned on their head is something you dont forget in a hurry!

*snip*

Well, from what I have read, EAX is just some fancy reverb effects unit - if thats what you want your sounds to go through, then so be it Maybe games designers use that so they dont have to spend quite so much time making things sound good and thats why it sounds better.

Im not clued up on it enough to enter into a detailed discussion, I am not a gamer and so that side of reproduction doesnt bother me. I will stand by my assertion that a proper surround sound receiver is significantly better than any 7.1 PC speaker rigup though

 

[fish99]

When I converted the MP3s on my Sony NW from 320kbps to 192 kbps to fit more tracks on, there was a discernable drop it quality across the board, every track. It was relatively slight, but I know all that music very well and I could tell the difference. 128 kbps MP3s sound like junk to me. If people can't tell the difference, that tells you more about the people than the sound. Most people will tell you their £100 Alba midi system from argos sounds 'fantastic'

I have tested and couldn't tell the difference between CD and 320 kbps MP3 though (on grado phones connected to Audigy 2).

 

[james.miller]

When I converted the MP3s on my Sony NW from 320kbps to 192 kbps to fit more tracks on, there was a discernable drop it quality across the board, every track. It was relatively slight, but I know all that music very well and I could tell the difference. 128 kbps MP3s sound like junk to me. If people can't tell the difference, that tells you more about the people than the sound. Most people will tell you their £100 Alba midi system from argos sounds 'fantastic'

I have tested and couldn't tell the difference between CD and 320 kbps MP3 though (on grado phones connected to Audigy 2).

that's because your re-encoding a lossy format. there is always a drop in quality. It's better to encode from the original @ 192k than it is to re-encoded 320k to 192k. If you know your music that well, i'd invite you to have a go at a DRZ blind test special

 

[DRZ]

that's because your re-encoding a lossy format. there is always a drop in quality. It's better to encode from the original @ 192k than it is to re-encoded 320k to 192k. If you know your music that well, i'd invite you to have a go at a DRZ blind test special

Im not sure I still have the ISOs any more, but I would be more than happy to go through the whole process again to create one, it only takes a couple of hours

 

[TooNice]

Nope - the way blind tests work are that there is a requirement that the subject gets everything right for it to be considered reliable.[...]

You misunderstood what I said there.

When I quoted you, I bolded the terms [WMA lossless]. You followed up by saying that people favorered WMA to other format (which I assume you mean other lossy format).. If that is the case, then it suggest people can tell the difference between lossy and lossless.

However, I would definitely fail the ABX test between FLAC and WAV because there is not difference. If what you did is compare lossless (WMA Lossless) to another lossless (WAV) then all you proved is that people use their pinky to guess when they are not sure if there is a difference. Not the most exciting findings IMO. I still think it is more interesting to simply compare the various lossy format to a lossless format to see how good the lossy format really are.

 

[DRZ]

You misunderstood what I said there.

When I quoted you, I bolded the terms [WMA lossless]. You followed up by saying that people favorered WMA to other format (which I assume you mean other lossy format).. If that is the case, then it suggest people can tell the difference between lossy and lossless.

However, I would definitely fail the ABX test between FLAC and WAV because there is not difference. If what you did is compare lossless (WMA Lossless) to another lossless (WAV) then all you proved is that people use their pinky to guess when they are not sure if there is a difference. Not the most exciting findings IMO. I still think it is more interesting to simply compare the various lossy format to a lossless format to see how good the lossy format really are.

What I was trying to get across but obviously didnt is that people thought that the original CD track was compressed and that the WMA lossless file was the original track.

Hence proving they didnt have a clue.

 

[TooNice]

Well, my reply to the 2nd part stands then. Just replace "WAV" with "CD Track".

You've shown that when cornered, people resort to guesswork. It is not something I consider particularly surprising.

What I would find more interesting, is how often people would be able to tell a difference CD track from a 128kps out of 20-30 trial. CD track from APS setting. CD track from APE setting. APS and APE setting. etc.

Obviously, someone who get half right is clearly relying on luck. But someone who get 18/20 probably is hearing a difference, and the 2 off could be due to fatigue and other reasons. Foobar take this into account when you do the ABX test.

 

[DRZ]

Well, my reply to the 2nd part stands then. Just replace "WAV" with "CD Track".

You've shown that when cornered, people resort to guesswork. It is not something I consider particularly surprising nor interesting.

What I would find more interesting, is how often people would be able to tell a difference CD track from a 128kps out of 20-30 trial. CD track from APS setting. CD track from APE setting. APS and APE setting. etc.

Obviously, someone who get half right is clearly relying on luck. But someone who get 18/20 probably is hearing a difference, and the 2 off could be due to fatigue and other reasons. Foobar take this into account when you do the ABX test.

Im not sure I follow you - there was a 128k track, a 96k track etc on the CD image I made. I did the test three times, all with different types of music.

Are you trying to say that when these people got it wrong (when they couldnt tell me which was better than the other) they were probably hearing a difference but just... wrote down the wrong answer? If there was a difference, a clear and discernable difference of which they were absolutely sure (see Tommy B's post above) then why cant people get the same difference right twice in a row?

 

[james.miller]

Well, my reply to the 2nd part stands then. Just replace "WAV" with "CD Track".

You've shown that when cornered, people resort to guesswork. It is not something I consider particularly surprising.

no, what he has shown is that people thought the wma lossless was the original while they thought the real original was one of the compress samples. In other words, they rated the WMA lossless above the original. Proving without a doubt they dont know what they are looking for if they rate two theoreticallt identical samples differently.

 

[fish99]

that's because your re-encoding a lossy format. there is always a drop in quality. It's better to encode from the original @ 192k than it is to re-encoded 320k to 192k.

That's possible I suppose. Actually, thinking back, about two thirds of the music was freshly ripped from CD to 192 kbps and about a third was coverted from 320 kbps MP3s.

 

[TooNice]

I am pretty sure I understood the experiment and result, but I am interpretating the conclusions differently. In one of the test, you made people compare a lossless file (WMA) and a CD Track (Source) right? The right answer here is: they are the same. But most people rated one higher than the other, and it turned out WMA took the cake in this instance.

Your conclusion: people obviously don't have a clue, since they think there is a difference between two identical track.

My conclusion: Of course people would think there is a difference, since they are expecting one. It doesn't mean that people actually think there is a difference, it merely mean that when people are forced to make a choice, then they will flip a coin.

If I have the time, I can get some friends over and make them compare two track encoded at 32kps with one being a copy of the other. I am willing to bet that once I get the result back, I will of course, get a similar result as you: perhaps "test.mp3" would end up being picked as the higher quality track, perhaps "a copy of test.mp3" would end up being picked as the higher quality track.

Since they (hypothetically) can not tell the difference between two identical track that are of crap quality, can I come to the conclusion that people obviously don't know what they are looking for, and hence won't be able to hear the difference between 32kps and lossless?

I will assume that the other part of the experiment do include tests between 96kps and lossless, 128kps and lossless etc. I won't critique those much, although three runs isn't much, and again, tracks you are unfamiliar with will be harder to discern. I won't say that the difference is black and white.. but I do find it ironic that someone who advocate "purity" in music, would seem to advocate not staying true the source to the files. Or perhaps that is not what you are saying, but then, I don't see the purpose of the experiment (other than to use it to discredit people who claim they hear a difference, who may or may not - I don't believe no one in the whole world can't, and I don't believe there aren't a few who can even in this forum).