Greetings,

This is my first post, so be gentle with me.

I am curious if anyone in this forum has done listening tests with different lossless codecs using high resolution 24/96, 24/192 pcm files? What were your findings?

I am also very curious about any comparisons that have been made between the DTS lossless codec and the Dolby True HD (MLP) lossless codec. Anyone have any listening experience with these puppies?

Has anyone actually listened to a download from MusicGiants? They max out a 1080 kbps and claim "lossless", CD's are 1480 kbps. What am I missing here?

Thank you,

JW

Doing listening tests for lossless codecs is a futile idea. If there were any differences it would most likely have to do with the decoder and not the file itself. Lossless files are identical to the original waveforms when decoded.

I don't know about DTS v.s. MLP, but I'm sure someone here might be able to give you a rundown or you could try a google search on the subject.

MusicGiants offers WMA DRM Enabled Lossless files. The files are not all 1080kbps, that's probably an average or something. When they say they are lossless they mean that only unnecessary information is discarded. All of the original coded file can be rebuilt without losing information from a decoded lossless file. And IIRC Redbook CD Audio is 1411kbps.

IMHO, FLAC lacks detail and has a narrow soundstage. APE is better, but many find it's sound excessively warm. Wavpack is by far the best, it's sound is highly detailed, the blackness between notes reproduced with razor sharp detail, the air surrounding the instruments filling the room, or something!



Oh, and you might want to read TOS# 8...

And if you simply don't understand what "lossless" means, try http://en.wikipedia.org/wiki/Lossless

I think the OP meant something like abxing 24/96 vs 16/44..

People have a hard time ABXing at 128 kbps, so I doubt you will receive any results for 24/96 vs 16/44.

Yes, in fact I was just reading about that. Both companies are trying very hard to tator to audiophiles, while enforcing watermarking to the nth degree. I don't see what the difference is? they are both lossless does it matter? if we are discussing the objective differences on a technical level then maybe, other than that

zen-infinity> you should take a look on this.

.

If people have trouble (i.e. failed) to ABX 13 bit vs 16 bit file, it's unlikely that they can perform a positive comparison between 16 bit and 24 bit material.

For high frequencies or sampling rate, the situation is even worse: 22050 Hz contents seem to be perfect to all people.

You know what irony means, don't you...?
And try using preview function before actual posting.

i believe sarcasm is a more appropriate term.

Thanks for the preview clue in. doh!!

I think the irony here is that this is a listening forum and it seems that no one can actually hear a difference between anything. "Bits is bits", and according to the posts above, all lossless algorithms are created equal. Which may be true, but in my experience highly unlikely. The assumption that all lossless codecs are created equal is logical in theory. But, without critical listening tests and comparison, at this point, it is just an opinion of someone with no actual reference point. I realize that to some of you this game of inches seems futile, but it is where I live and work. So if you think I am being nit picky or just plain insane, please excuse me.

There also seems to be a lot of hostility towards people that can actually hear a difference. Why is that? Why the hostility? Just because you cannot hear differences doesn't mean they do not exist. Just because you can't hear air around instruments or a black background or more significantly a difference between 128kbs codecs doesn't mean that everyone is going to hear it the same way you do. Hey, if your ok with 128kbs, god bless you. But to take swipes a someone because they think differently is unproductive.

There is no doubt that all the posters on this thread have strong opinions about lossless codecs, yet no one has actually LISTENED or has made comparisons. What gives? What qualifies you to have such a strong opinion when you haven't actually listened or compared lossless codecs?

I work with DSD and PCM on a daily basis, surround and 2 ch.(and vinyl for that matter) There is a subtle yet remarkable difference between a DSD file and the same file downsampled to 16/44.1 (level matched within .5 db, both files running at realtime through the exact same signal chain w/ realtime A/B switching). Without having a trained ear a lot people cannot hear a difference However, you just need to know what to listen for and more important you need a system that has high enough resolution in order to identify subtle changes. Once a clue was given on what to listen for, those same folks who couldn't hear a difference (DSD vs. 16/44.1 ABX), now could because they knew what to focus on, it's a matter of training your ears. Your hearing can be easily tricked, it takes time and experience to train your ears on what to listen for and how to listen critically. Once you have done this for years differences become more easily apparent and music with artifacts casued by processing whether it be downsampling, compression, eq, limiting, reverb, etc become more irratating because you brain and your ear has made the connection before. Does that make sense?

In my expereince evrything makes a difference. Whether you care about it or whether it makes a difference to you or not is another matter.

The thing is that lossyless is just that, no loss. This is how it is done:
A track is ripped to wav, lets call it file1.wav. Then you create e.g a FLAC of that, lets call that file file1.flac. Then you extract that file to file2.wav. Then you compare file1.wav and file2.wav against each other. Not listen to them, but binary compare them, see if they are equal. If they are equal (and remember we are workiing with 0's and 1's here, it is not hard to determine if they are equal, and equal here means 100% equal) then you can listen as much to them as you want, and hear as much difference as you want, but they are equal.
And FLAC is througly tested, it is lossyless.

No one with half a brain cell would even waste their time on lossless listening tests.

Apples and oranges my friend, you claim that if you have DSD and then downsample it, you hear a difference, that's fine because you edited the file. Lossless compression is not editing (lossy you could count as edit, since it throws bits away), lossless is just storage, if you trully work in audio field you should know what a wave form is, and if you do you should realise that having two files with identical waveform (which is what having uncompressed and lossless compressed files is) it is a physicall impossibility to have different sound come out on playback of each file.

Yeah, my post was intended to be just a tad sarcastic.

The only reason I can think of for there to be any difference at all, even if it was so small as to be virtually unmeasurable on a good quality system, would be that a losslessly compressed file requires more CPU usage and less storage media reads which might in theory change the amount and type of noise from an analog audio source that's in the computer (like onboard audio or a soundcard.)

If you're using an external DAC then even that slim possibility of a difference is eliminated.

If you think that the losslessly compressed audio will have more "jitter" than the original then you should buy my new audiophile grade .wav dejitter utililty. It uses proprietary software techniques to carefully realign the bits in audio files stored on your hard drive so that they play back with minimum jitter when using the correct hardware.

I agree, the example of dsd vs 16/44.1 to an original wav file vs lossless packed and reassembled wav file is like comparing to apples and oranges. Point taken.

What I think I was trying to say is that some people can hear subtle changes at very high sample rates and bit depths. Perhaps suggesting that other tests mentioned above regarding 128 kbs and 13 bit were flawed. Just because you didn't hear it doesn't mean it didn't happen. Again training, experience and ear brain connection.

Regarding lossless, there is a lot of theory posted here but no experience in practice. I realize comparing lossless codecs could prove to be a wild goose chase. But quite simply, I am curious and don't take things for granted, especially any "process" that has to do with audio. Very rarely do you get something for nothing. I strongly feel theories taken for gospel should be challenged and examined. You just never know.

What I do know is that there are very few ways to improve signal quality and infinite ways to screw it up. With my half a brain cell I posed this question to the forum in earnest.

One last try. How about DRM? Has anyone ever compared different DRM schemes for sonics? Ok, before you go off and freak out about how anyone would want to drm thier files........I agree with you, it sucks. I am not talking about this philosophically, I just want to know if anyone has ever LISTENED to different drm schemes and has notated any sonic artifacts?

Did someone said that?
It may happen. Some people are maybe skilled enough to hear the impact of a very high sampling rate or an unusual bit-depth. BTW, a lot of people are claiming that they could hear such difference -- easily! But on HA.org proofs for such claims are needed. It's a rule of this board. People have to differenciate these HR files on double-blind test conditions.


I believe you're wrong. It is true that here are no listening tests published on organized on this board. But it doesn't mean that no experience are done.



Yes, this is true. But there are hundreds members of this board which are daily using different lossless encodings and listening to these files. And for years now. If there were any form of degradation with lossless files, some reports would be posted. If peoople don't question anymore lossless, it's because it's safe, objectively and subjectively. Experience is confirming the theory, and bit-to-bit comparators are here to prove and/or confirm that no coding errors are introduced. BTW, you're free to investigate, experiment, review... lossless encodings. Just keep in mind that experience with audio quality is most often useless if it's not performed on blind conditions.

So what? Does it make sense to suspect a lossless processing to introduce any form of loss, just because other form of process are lossy? What about listening then? Did quality progressively degrade by listening digital files? Why not? If you want to be suspicious, you can suspect everything.

You aren't getting something for nothing with lossless. The lower the bitrate, generally speaking, the higher the complexity of the audio coding/decoding. This is relative to the different codecs, not the complexity of the music. In order to be called a lossless codec, the code/decode process has to be mathematically lossless.



You won't find anyone here selling snake oil if that's what you're asking. DRM is handled by the CPU the same way the decoding of the audio is. In order for DRM to affect the music, applying the DRM would have to change the music in a certain way. DRM by it's nature just protects the file from being illegally copied etc. It should in no way change the sound of the file. I have never heard of this and it sounds crazy to me to be honest with you.

I don't know, much of the information contained in the hydrogenaudio forums/wiki is based on logic. The questions you are asking seem to throw logic out the window. Asking if DRM schemes change the way music sounds is kind of like asking if having a different artist tag on the file will change the way it sounds. It shouldn't matter is the point I'm making. If DRM changes the way the file sounds, the DRM is fundamentally flawed IMO.

To be honest I am just starting to educate myself about drm. I was operating under the assumption that it was embedded in the audio signal, similar to scms, which definetly has a sound.

Seriously though, you should probably read the posting guidelines specifically
ToS# 8.

Without proof, that claim is an objective opinion, and should be stated as such.

End of story. Encode your wavs with a lot of different lossless codecs. Decode them and compare each one with your original files. They are equal - if not, something went wrong - and by no loss of information in every encode you can understand that a lossless file will not have better 'sound' than another.

I dont need to test with my ears that are greatly affected by subjetive manners, even stress, when I have a 'math way' of doing this just comparing bits.

Given that there are so many other lossless compression techniques used in different digital applications (ZIP, RAR, PNG, etc.), I don't know why you would find lossless audio compression so suspect. It's just data, after all.

Unfortunately, such kind of noise is highly random and irreproducible. The noise from analog signal generator mostly comes from power line noise, which in turn depends on the noise of power supply, ambient temperature, and high-ampere equipments (e.g. CD player, HD player, etc).

This leaves only CPU usage to account for. If CPU is working too hard then decoding may be interrupted and the playback buffer becomes depleted, causing audible click. However, with the current crop of processors, it is very hard to reach this situation. In my case, if the processor ever maxed out, usually I am doing something that causes me to ignore the sound quality completely (e.g. playing F.E.A.R )



Comparing lossless not only a wild goose chase, but a futile attempt. As lossless codecs will faithfully reproduce the original waveform bit-by-bit, it is plain impossible for our ear to hear any difference. Any perceived difference therefore is caused by psychological distractions, e.g. whether in the last 5 minutes I was kissed or slapped by my girlfriend ...

I completely understand and respect your stance. I just thought this forum was about "Listening" Tests. I thought it would be interesting and perhaps educational to hear from someone that actually had experience and expertise testing and listening to different lossless codecs, my mistake.

In the spirit of fun I have riddle for you wiz kids, it's a little tricky. This happened to me and a colleague not to long ago, and it's very much on topic.

You send a replication plant a 1630 U matic tape and AIT for stereo hybrid SACD replication. They burn it to a glass mastera PERFECT match bit for bit!!! Yahoo!! Everything is PERFECTOMUNDO! Bit for Bit verified! End of story! Let’s go press up 10k units of that puppy!!! Right? You done now, bit for bit verified says it all, go cut that PO for 10k units right? In Richard Dawson voice, "Survey says!" (short pause for drama) "BuzzzzzzzznnnT!" (ok that's my lame buzzer sound, back to Richard). "Ok I am sorry, we have some nice matching luggage for you and a years supply of Rice-A Roni." NEXT! Based on your bit for bit theory, you would have just purchased $25k in defective product. That's a lot for drink coasters!

However, you know better and that there are infinite ways for things to go wrong during a transfer process and that bit for bit verification is a wonderful tool but not the gospel. You get the test disc and you LISTEN to it. 1st you qc theDSD layer which passes muster in listening (damn, there's that word again) tests when compared to original files on the hard drive through the same d/a in real time.

Then you get to the CD layer. There is music coming out of the speakers yet the life is sucked out of it. Huh? This can't be the difference between DSD and redbook PCM? Am I going insane? You listen to the redbook files on your DAW. Sounds right! CD sounds rolled off in comparison. Huh? Next you take out your backup clone of the U matic you sent to the plant and compare it to the CD layer in real time and its obvious there is more top end, definition, openess and air. You go through several transports and D/A combinations and it still sounds rolled off. But hey it's bit for bit verified right!? Hmmm.

You call the plant and say what the hell happened!? They treat you like you are insane, "because bit for bit these are exactly the same as your 1630". You ask, did anyone actually listen to it over there? "Yes", they say and "we will do it again and be extra careful this time, we'll even use headphones". You explain to them through actual listening tests your findings, "It sounds like the top end is rolled off". We dumbed it down further and said, "It’s kind of like if someone put a bed sheet over your speakers, things sound a little muffled".

The report comes back a day later and says, "original master source material old, worn and dated, all data verified bit for bit". What? That's it? Did anyone actually listen? Did they compare it to the SACD layer? The plant confirmed, "Someone listened to it carefully per your request and found no abnormalities, and besides, stupid, it was verified bit for bit." ok I took some poetic license with that one. ; )

Basically, the plant takes the "bit for bit" stance and says you are crazy. In a nut shell, go jump in a lake. "We have very expensive high tech machines and our technicians have verified that the files are identical." You go back with your colleague and LISTEN again. NO F-ing way! It's not right! You call back, raise holy hell and demand to go up the food chain, you want the THE authority on glass mastering. You draw the 20 year veteran, Mr. expertise-o extraordinaire. You know what you are about to tell him will make you sound crazy. But you know your ears, your colleague is a 20 year mastering veteran and you've mastered and recorded for over decade and been in the audio biz for 24 years. You both know what your hearing or perhaps...could it be? You are ......INSANE! TA DA DA DUM!

And now a word form our sponsor.

I am going to leave it here for now. I'm pooped and obviously feeling a little wacky.

So bottom line. "Bit for Bit" verification wise, the umatic tape at the plant, the glass master and the CD layer all check out PERFECTO. And these two wild and crazy guys who actually listen, claim they are hearing "THINGS". TA DA DA DUM!

What happened?

Just because you didn't hear it, doesn't mean it didn't happen.

Try to keep an open mind, especially when doing listening tests. I agree in many respects audio reproduction is a science, but it cannot be overlooked that it is also an art. Especially, what is becoming the lost art of listening. Like anything else it takes training and experimentation in order to gain experience and eventually expertise. It's within eveyones grasp, if they are will to take the time to do it right.

Part way down the page under "History"

http://www.lipinskisound.com/about.html

.. a lack of substantiated testing as opposed to accepting someones word.

ABX testing

I'm not one for repeating what's already been said but, if there's a difference between lossless codecs, something is broken in the playback chain.

Sorry, its up to you prove it did happen. Otherwise you get all sorts of unverifiable blue elephant theories.

If bit-by-bit comparison is not good enough for you, then you actually are claiming that a file is never the same twice. (since a file is always bit-identical with itself). If that is the case, any listening test would be unusable as results would never be reproducable.


That's just being romantical, and unfortunately not really to the point here. Does an English book report about George Orwell's 1984 written in Word change into a report about Shakespeare's Hamlet after zipping and unzipping it? You might not have seen it, but that doesn't mean it isn't possible, right?

Of course not. But we're in a world ruled by Intel and AMD, not by Gandalf and Harry Potter. If files are bit-to-bit perfect, most people won't see any reason to worry with listening investigation about possible but irrational difference.

But as I said, you're free to investigate and perform your own listening tests between lossless encoders, even if it sounds useless to most people. And when this task will be over, and if you're still interested by possible sound degradation, I've other very useful suggestions:

- which hard drive brands produce the best quality? Some friends told me that sound is warmer with Hitachi, but that MP3 have more soundstage with Maxtor.
- the influence of computer's supply on stereophonic image. Is sound better with 300W or should we switch to 480W power supply? Your experience is welcome.
- AMD or Intel for better bass response?
- which CD-Rom brand for better extraction? Are really Plextor giving a better accuracy to high frequencies content?
- and what IDE cable are the warmest?

A sceptical person, interested to compare lossless files sound quality, should test every components of his computer.

I do agree with guruboolez. Let us put several things straight:

Bit-by-bit perfect means just that: The copy is exactly the same.

Does it sound the same? Yes, it should.

So why doesn't they sound the same? Let's see here:

1. Different decoders. Especially with lossy codecs, different decoders will cause different waveform reproduction. Some good, some bad. With lossless codecs, the reproduced waveform must be the same. If not, then the decoder is buggy.

2. Different sound reproduction equipment. I do believe that Umatic tape player and CD player are 2 different equipments. Most likely both use different audio waveform generation chips, and thus will produce differing audio signals. Not to mention the signal path taken through the output circuitry of both players.

If we are rigorous in testing, ensuring that every darn sound reproduction equipment is exactly the same, and also ensuring that the playback environment (e.g. power supply, ambient noise, psychological mood, etc) then 2 bit-by-bit identical WAV files must sound the same.

Ambient noise we can control. So can we control power supply. The one thing we cannot control is the psychological mood.

So, how can we ensure fair and objective test? By using an anechoic chamber with high-fidelity mics to re-record the output of the speakers, as WAV files. Also allow several minutes between 2 WAV files to let the chamber settle.

We repeat the test the next day at the exact same time, but this time we reverse the WAV sequence.

Then we compare the result bit-by-bit but with tolerance to allow for noise in the cables.

Outrageous? Yes. But in my opinion, it is the only way to conclusively proof the difference between lossless codecs.

And I think I already know the result.

At least on AMD X2, audiophiles should avoid running two lossless encodings at the same time, because there will be quite high intermodulation distorsion between them..

AMD is working on a new CPU stepping with better shielding between cores, to reduce the interference.

For the time being, better use lower frequency CPUs (up to 400 MHz) for the best sound..

In other news, cheap SATA cables have been reported to give a subtle, but very disturbing metallic timbre to the sound..

Ok. Normaly to make double-blinded comparisons the two files are decoded to random wavs and them you listen trying to find diferences. But in the folder you have two identical wav files, with only diference is the name. I think that is simple too dumb to me try to find diferences between two identical files/waveforms.... But you is free to do what you want. You can even try to start an listening test, but I doubt you would have many voluntary and I doubt even more that they can find diferences between the files.

no shit?

Yes, and it will use water gates instead of transistors, thus removing all electrical interference. The basic material was discovered in deep sea in the 80's by Commandant Cousteau and his team.

I'm testing an early prototype of it in my bathroom; the only problem is when a calculation overflows, the water overflows too

I feel tired.

Actually, CD is 1411.2 kbps / 1411200 bps.

which is obtained by multiplying:
44.1 [kHz] * 16 [bit] * 2 = 1.4112 [Mbit/s]

man, i'm always taking things too seriously

Where's zen-infinity hiding now? Any more arguments to distract us from 'science' ?

@Shade[ST]

bad for him we don't have 1990 anymore. i could hear my hd through the speakers.