Can you hear the difference between lossy and lossless audio?


#1

Recently, on another forum, I was in a heated debate with someone who was trying to tell me that it was impossible to hear a difference between lossy 320kb/s MP3 and lossless audio formats such as FLAC. While googling for a quick and dirty blind test to be sure I wasn’t just talking out of my ass (as I told him I was sure I could based on my system, but not from real world tests), I found this:

(once you click on the link you need to scroll down past the video playing)

https://www.theverge.com/2017/4/5/15168340/lossless-audio-music-compression-test-spotify-hi-fi-tidal

It was relatively easy for me to hear the difference between the three tracks. The 160kb/s was obvious to me each time. I will admit if I hadn’t been listening particularly close, the 320kb/s could have been mistaken for lossless audio. But I was listening close and with only a second play-through after figuring out which was the 160kb/s file, it was obvious to me which was which. The lossless audio simply sounded cleaner with faster transients and it just sounded more tonally correct. It really wasn’t guessing. I was certain every time I chose a track out of the three. He thinks I’m lying to him and that I was guessing.

My argument with him was that yes, sure, most people probably can’t hear a difference based on their hearing acuity and/or speaker setup (or just plain old headphones), but with the right system and fresh ears, the differences should be easy to tell and they were. Obviously he falls into the former category, not the latter. That’s fine, but I just take offense when people try and make it seem as if I were crazy for spending so much on audio equipment.

Can you guys hear the differences easily too? I don’t have crazy expensive speakers like many here do, but they’re most definitely resolving enough for such a test.

Edit:

I do want to add that they were using MP3 for the lossy encoding. Services like Spotify who offer lossy 320kb/s audio, encoded with a much more efficient encoder known as OGG. Other streaming services use AAC which is also a step up over MP3 at the same bitrate. This means with the same 320kb/s information, more audio information from the lossless source can be kept. I would like to set up a test to see if I can hear the difference between FLAC and OGG @320kb/s too. I think I’d still hear the difference but the gap would narrow even more.


#2

I wanted to say, “Absolutely, there’s a difference!”, but of course - as you point out, it’s not absolute. It’s system and hearing and training dependent. I think there is some level at which it’s provable, both with measurement and DBT. But it’s really easy to make a test that goes either way.

I would hazard that the vast majority of people who have a way of listening to music could not tell in most cases due to one or a combination of those things. Plus, there’s the whole not caring or not wanting to hear one thing or the other… ; )

When you start trying to test it, it gets murkier, since you’re often comparing different things. What is doing the decoding - software/hardware? Are all of the codecs being decoded in the same way/by the same chain? Is that the “best” means of decoding that particular codec? Etc. …at which point my eyes glaze over and I lose interest. I don’t really have any desire to prove it to anyone.

But I also think there’s often a difference between lossless and the .aiff it was made from, so… itwasntme_gif


#3
badbeef said

But I also think there’s often a difference between lossless and the .aiff it was made from, so… itwasntme_gif


Indeed it is at least in my setup. But as I was discuccing with Ted, he suggested that this can also be system dependent (with Bridge different than with USB connected devices, depending on revealinngness of system, important to exclude other circumstances that would make differences etc. etc.) So everyone has to try for himself…in my case compressed flac was the worst, next came uncompressed flac, than again noticable better aif and only a tiny bit better in bass (could be imagination only) wav, but without metadata, so unusable for me.

#4

Yeah - lossless vs. the source is another one of those things that can get folks going. “It’s LOSSLESS!!! It CAN’T sound DIFFERENT!!”

So it often gets back to the issue of - is it possible to make a comparison that is meaningful? The Lossless has to be decoded (in my mind - correct me if I’m wrong) in a way that the .aiff doesn’t. But typically, that right there means there are two different processes going on in the chain.

Usually, for me, (at least in the ancient past when I formed this calcified opinion) this was because all CD players “natively” decode CDs, but if you’re playing say, FLAC or ALAC, you needed to use a computer, or some sort of software or hardware decoder that is something other than the device unpacking the CD. Which right there says, “totally different chains, and as such, perhaps not the most meaningful comparison”.

One example I have is that for 15 years, I’ve had Squeezeboxes, have my whole library ripped to HD (but kept the CDs), and at one point got a Transporter (which is still a pretty nifty bit ‘o kit, at least from a historical perspective) and noticed that FLACs sounded different from the rip of the CD - and also from the CD playing on a transport.


#5

For those who truly want to know what they can actually hear, Foobar offers an ABX Comparator which provides a true double-blind test. You simply designate the two files you wish to compare, making certain the levels are matched. Keep in mind you need to correctly identify the higher resolution file 23 out of 24 trials for a 95 percent confidence level, the typical standard for statistical significance. If there truly is a discernable difference, this will be easy.

I continue to find telling the degree to which audiophiles often insist one needs an excellent system, careful setup, a great deal of listening experience, the right source files, plenty of sleep, etc. to reliably discern the difference between lossless and data compressed files.

If we find this test difficult, our claims high resolution files sound better than Redbook appear ludicrous. Even more difficult are assertions various file formats sound different.


#6

Elk - I take it you feel/know it doesn’t make a difference? Is there a level of compression that you feel does?


#7

I would agree that a better system helps identify the differences, after all extracting the most information from the source is one of the goals of a high end system. The real determinant here, however, is the listener. Regardless of what some may say we all hear in our own way. A most telling example of this is how new firmware for the DS is received, some love, some like, some dislike, and some hate (or at least that’s what it seems like). Personally I have tried comparing compressed files with uncompressed and was able to hear differences. The problem for me was identifying which was which. I was not very good at it. I need more practice listening!


#8
pmotz said Regardless of what some may say we all hear in our own way.
At a minimum, we each have preferences and value different aspects of reproduced sound. For me, it is timbre and microdynamics.

#9

I’m currently listening to an old MCA CD I unearthed of The Best of Steely Dan, which has a great selection of tunes, but which has a sound reminiscent of not-so-great vinyl record playback (rolled off top, muddy mids). I just thought of this because of Paul’s recent Post. I could play a track from this, and the same track from one of my clean copies on vinyl, and I would bet most people would think the vinyl was the CD (barring vinyl tells, like a pop or tick).

Of course I have no idea if they’re from the same source, or why this was mastered poorly, etc. Not that all Dan is HiFi. Much of the older stuff is an embarrasment to the Donald ; )

I was recently watching the Classic Albums DVD about the making of Aja, and at one point, with Donald and Walter (RIP) sitting at the board playing back tunes and isolating tracks, Donald quips, “This was the period when I was singing like Jerry Lewis” (or words to that effect).


#10

Beef, in another thread I told about differences between Bill Evans/Another time vinyl and Redbook file…this was also an example where regarding top end performance I would have sweared vinyl is CD vice versa :wink: A lot we hear is mastering.

Regarding those small differences of flac compressed/uncompressed, flac uncompressed/aif etc.:

At times when I don’t have the patience to do such comparisons or hear only little difference I apply multiple of such changes that should be better but are hard to identify.

At least the sum of them usually make a noticeable difference at the end. If they don’t I know what to pass on.

But: things change when systems improve…so a sporadic recheck makes sense.


#11
badbeef said Elk - I take it you feel/know it doesn’t make a difference? Is there a level of compression that you feel does?
I have played with the Foobar ABX comparator with 320kbps files sourced from an excellent Redbook recording. When listening double blind the MP3 files are much harder to pick out than I expected.

My experience is audiophiles have strong opinions and condemn that without sufficient pedigree, but when it is time to put money where your ears are they have many, many excuses. If even high bit rate MP3 files suck digital rocks, they should be easy to pick out. Yet, few audiophiles can.

I have not reached a definitive conclusion as to when higher resolution exhibits diminishing returns.

As always, the skill of the recording engineer and the polishing efforts of the mastering engineer are vastly more important.


#12

Recently I invited an old friend over to hear my system, and to listen to some music. We have been friends for 54 years, and in the late '60s and '70s he always had the best system. While I continued my musical journey, going to hundreds of concerts and upgrading my system he sort of lost interest, but got into home theater. He put together a decent system, but hasn’t upgraded other than his HDTV and an Oppo.

Since he has been over, I upgraded my preamp, and DAC [3 times]. Plus better setup and added the dedicated computer for hi-rez and now Tidal. So I said you really should hear it, so much better. His reply was that he probably wouldn’t hear the difference, as it sounded great last time he heard it.

The point of this story, is that we listen, really listen. I know he won’t be able to remember how my system sounded four years ago, but the difference to anyone posting would be that of a much better overall sound. I know when he comes over he will enjoy it, maybe come away thinking it is the best system he has ever heard in someone’s house. But it won’t be a revelation, or motivation to improve what he has.

So debating sound with people who feel the need to tell you that you are fooling yourself, that have made up their mind, aren’t going to hear any differences, whether or not there are any. The last time I did any careful evaluation type of listening was a comparison of the DMP and the transport in an AR CD9. We both heard the same things, and at the same time felt that the differences where slight enough that we would be happy with either setup.

When it comes to comparing codecs, I think the differences become more obvious when I listen to whole albums or concerts. The Grateful Dead give away 30 downloads each November, they are MP3s. For casual listening they are fine, but if you listen without distractions, you will notice that something is missing. Would I hear it in an A-B test, I’m not sure. And I have gotten to the point where I don’t worry about it. I just try to listen to the best available version available.


#13
jeffstarr said Would I hear it in an A-B test, I'm not sure.
You should if you can other wise tell. Remember, an ABX can be as long or as short as you would like.

#14

It’s recently been suggested to me that I can sound condescending (my wife would’ve agreed), so apologies in advance:

I wonder about what you’re getting in terms of reference using a piece of freeware on a computer as an ABX tester. How resolving is it? Is it, in effect, “narrowing the bandwidth” of everything you are comparing on it? This is another example of why I’ve come to feel that “comparisons” of various things are often fraught and rarely meaningful, when you really get down to it. Unless there is some compelling reason for it, the endless arguments over tiny differences make me feel my time might be better spent on something else.

To me, these comparison questions are interesting from the broader perspective of, “What is the question, why do you want/need to know, and is the test for it a good one - able to resolve the differences, if any, and giving a meaningful answer to the question?” Sorry, I’ll own up to bias - too many years working for clients in marketing research, where millions are spent gathering and parsing meaningless data on meaningless questions.

I will admit to another bias, as I’ve never seen any need for compressed (or additionally compressed, I guess I should say, since any codec is compression) versions of music that I have. As a result, I can only fit several albums on my small iPod, but I never felt the need to carry my whole collection around with me. I just cycle the contents from time to time, and mostly only use an iPod in the noisy environment of the car or plane anyways. I just don’t even like the IDEA of throwing away data or changing the original file because someone else thinks I don’t need it.

Then there’s having something like the DMP/DSJ, which can play SACDs “natively”. Not sure if it’s capable of playing mp3 files. I don’t have any, and don’t care if I can tell 320kbps from 321. But that strikes me as a better sort of reference for hearing the differences - except, of course, it’s 20x DSD upsampling everything, so…again, not meaningful! : )


#15

There are of course ways to use Foobar2000’s ABX tester without actually using it to decode and render the audio. It has a UPnP output option which would send the audio file to a UPnP device like the Bridge II. Other people who don’t own a Bridge II could use a reference UPnP network audio renderer from the likes of Melco, Aurender or Sonore (and many others). There are plenty of ways to use a computer without it having an affect on the audio. This way you get best of both worlds. And yes, you preface your post saying you can come off as condescending and I do get that sense from your last post. The DMP is a great product for what it’s built to do, but it has plenty of impracticalities in the context of what’s out there in 2018. To be more specific, it’s a disc/local player only in 2018. It may be “reference” in what it does, but it is a very limited product in the grand scheme of things. There are several other “reference” products similar to the DMP that offer Spotify Connect and native support for Tidal while also being a CD/SACD player.


#16

Seegs108,

I am not surprised with your experience on another forum discussing lossy vs lossless music files. For some reason, non audiophiles really hate the notion of high resolution audio and will bash it as much as possible. They seem to have this idea stuck in their head.

I had a similar experience almost 10 years ago on a Blu Ray forum and got in a heated exchange with a non audiophile. He was trying to cite a study by Myer and Moran (I think, if I remember correctly) about proving that there were no differences. But, their study was flawed. They compared a red book CD with a high resolution file that was down converted to 44.1/16. So, they were basically comparing a CD to a CD! And to top it off, the person I was arguing with had never heard a high resolution file and admitted it.

Since then, I have decided not to waste my time and engage with these people since we are not going to change their mind. I have never compared a 320kbs file to a non compressed file, so I really can not comment. But, I believe that music has enough nuances and detail that would get lost in compression so that it should never be compressed. My belief is so much that I do not have a portable device since I don’t want to listen to compressed music. Yes, I know that I am at the opposite extreme and will admit it.

So, it boils down to what you believe in and what makes you happy.


#17

Seegs - fair enough. Although I don’t know how “There are plenty of ways to use a computer without it having an affect on the audio. This way you get best of both worlds.” It seems to me that if there is a computer connected to the music chain in any way, it is having an effect. But that may be - I truly don’t know about that.

With regard to the comparatively limited-use DMP/DSJ, again fair enough, but I did mention that due to it being an upsampling device, it is no more of a meaningful comparison (if you are trying to determine in some absolute sense if different codecs are discernable from one another) than Foobar on a computer. It is a good way to play discs though, and many feel that it sounds better than that same disc’s files ripped and played back on an optimized computer chain through the same DAC.

I guess part of what I was trying to get at is, if you are not playing one format or the other back on something that is a good way of playing it back, you’re not hearing what it can do. Anybody who has a preference for a given format tends to have gear that optimizes playback of that format, whether it is vinyl, CD, files from HD, or whathaveyou. If that then requires different gear to get the most/best out of a given format, you can’t really say if one is better or that you can’t tell one from the other. If you are making comparisons on a chain that does certain formats better than others, you can only say that, on that chain, you can or can’t discern a difference.

What I’m getting at is that I don’t see how one can answer the question in your OP, except ‘IMS’ - in one’s system. Yet I agree with your OP.


#18

"It seems to me that if there is a computer connected to the music chain in any way, it is having an effect. "

I think you’re misunderstanding how UPnP works. The reason so many audiophile devices support it is because of what I mentioned before. You can use a PC or other non-audio-centric device as an interface, with beautiful GUIs and such, without having it affect the audio. This is because UPnP sends the file to the rendering device such as the Bridge II or the device I’m using. An analogy here to physical medium that you might have an easier time understanding would be to think of it as a way to literally give another device a CD, but in this case it’s a digital file. UPnP does nothing to the file other than send it over a network. The PC doesn’t decode, render our output anything other than the file as it is. It cannot have an affect on the audio because it’s literally not doing anything in the audio chain for playback. You could say that the PC is still connected to the rendering device through an ethernet cable so electrical noise may ensue, but a lot of people, myself included, use a fiber optic network adapter to galvanically isolate the network and PC from the rendering device. I can’t think of a way when set up like this that a PC could have any affect on sound quality. You’re literally taking the PC out of the equation like this and are only using it as a way to interact with what audio you want to listen to. At this point you should only be “hearing” the network audio render and DAC and how they affect sound quality. They too all sound different so careful consideration should be used .

The nice thing about a setup like this is that you have digitally downloaded files or ripped CDs and when setup and used correctly (galvanic isolation and UPnP) the CD should sound as good as if it were played back on something like a DMP given the network audio renderer has excellent hardware inside. Paul has discussed this before in some of his YouTube videos. The potential is there, but as he points out, it’s the method of delivery (like everything else) is what matters for the integrity of the music. Few devices meet the standards of the DMP so finding an equivalent delivery method for digital file playback with other devices give you few options. But devices from Aurender, Melco and Sonore really should be considered (and listened to) before you make such claims or give shade to other playback methods. Again, the potential is out there as Paul points out (And is his mission with the “big” server). But unfortunately the number of devices that meet the quality of something ala DMP are few and far inbetween.


#19

@seegs108 I am curious how a person that is using a NAS connected to a switch can add galvanic isolation if the only method allowed is a Cat 6 cable from BridgeII to the same switch. Do you need a special switch that allows for fiber optic cable connection? And how would that connect to the BridgeII with just the available ethernet jack?


#20
Deanhorn said

I am not surprised with your experience on another forum discussing lossy vs lossless music files. For some reason, non audiophiles really hate the notion of high resolution audio and will bash it as much as possible. They seem to have this idea stuck in their head.


It is precisely the same as audiophiles who, collectively, refuse to acknowledge that an MP3 or other data compressed file can sound superb, and/or are convinced that if a file has been compressed - even with a lossless format such as FLAC - is compromised.
badbeef said

. . . if you are not playing one format or the other back on something that is a good way of playing it back, you’re not hearing what it can do.


Very true. You need a playback chain with sufficient resolution and capability to compare any two formats.
Anybody who has a preference for a given format tends to have gear that optimizes playback of that format, whether it is vinyl, CD, files from HD, or whathaveyou. If that then requires different gear to get the most/best out of a given format, you can’t really say if one is better or that you can’t tell one from the other.
Yes, again (you are on a roll). In particular, I have met few who have spent any real money on vinyl who are not dogmatically convinced vinyl is inherently better than any form of digital. It becomes a theological position. There are, of course, those who reject vinyl as an instant knee-jerk reaction. Silly.

Along these same lines, I suspect many put together systems which reproduce the music they prefer and then buy music which sounds good on their system - narrowing their musical preferences even further. I find that a truly good system makes all types of music sound its best, but many feel their choice of music dictates what equipment they should purchase. Consider how many posts asking for recommendations are careful to detail what music they listen to. (Yes, if one has a very limited budget, taste in music can be highly relevant - if you listen only to lute recordings you need limited bass response can buy a less than a full range speaker.)