Lossy vs Lossless audio: Can you hear the difference?

[audacity]

People say silly things. One of the silly things that audiophiles say goes along the lines of "I don't want lossy audio. Yes I can tell the difference."

Well, yesterday there was a Lifehacker post on this very subject. They show a audio application (foobar2000) which added a ABX testing feature so that you could compare audio files between lossy and lossless formats, and it'll even keep score and see how you do!

I thought this would be educational to anyone who believes they have super-human hearing.

[cooper83]

I wonder if there's a flicker test too...

[JamesK]

I can certainly hear the difference between MP3 or OGG files and the original CD tracks. On some songs, the difference is significant. I often listen to MP3/OGG for casual listening, but if I want to really enjoy the music, I go back to the CD, often with headphones.

[57]

The test was at 320kbps and many magazines/testers have already stated that it's difficult to consistently hear a difference at those sampling rates. I have two comments.

Firstly, most people don't record at those rates.
Secondly, if you do use 320, there's not much difference in file size compared to lossless. Since HDD space is so inexpensive, I'd go with lossless.

[audacity]

I've done the same test at 192kbps and cannot hear a difference when compared to WAV files that were ripped using Exact Audio Copy.

My main point is that I've run into a number of people who said they could tell the difference between a WAV file and a 192kbps MP3. In the two cases where I challenged them to tell the difference (and they can choose the CD as the source), neither of them could tell the difference.

I'm pretty sure that almost nobody does this sort of back to back testing, and just assume they have a golden ear.

I can certainly hear the difference between MP3 or OGG files and the original CD tracks.

Did you do proper testing before you determined this? Or is that just based on faith?

[JamesK]

^^^^

It doesn't take a "golden ear" to hear the difference on some. For example, I have one recording, The Beatles "A Day In The Life", where there's one stretch where you can barely hear John Lennon's voice, whereas he's quite prominent on the CD. I have others that generally sound OK, but if you listen carefully, you can hear artifacts of compression, such as echo's spurious noise etc. The problem with lossy compression is that it removes what's supposed to be insignificant sounds. However, on occasion, those "insignificant" sounds may in fact be significant. Some new sounds that weren't in the original may also be created. Of course, there are also some people who are quite tolerant of distortion and can't hear what's plain to others. Years ago, I worked with a guy who couldn't hear power supply noise in his radio, yet I could hear it from quite a distance away. It was so bad you'd think he'd have to be deaf to not hear it.

As for comparing, the vast majority of my MP3 & OGG files are made from CDs that I own in my collection, so I'm very familiar with both compressed & uncompressed versions. I only listen to those files on my smart phone, either with headphones or through the stereo on my computer desk. In the living room, I listen to CDs exclusively, often with Sennheiser headphones and with the lights turned down. I suppose the best I could come to an A/B test would be to plug those headphones into my smartphone.

However, as I mentioned, compressed files are fine for casual listening. I just listened to a nice classical guitar piece by Sean Kelly and currently the "Lovin' Spoonful" are on with "Daydream". It's good enough while I'm at my computer.

[runnin']

Two points. Not all hearing is equal, just as not all eyesight is equal. So to assume that everyone hears exactly the same, or exactly the same as you, is a big leap. Second, a HTIB or entry level system may not be able to demonstrate the difference. All systems are not the same either.

[Jake]

My lossless encodes are typically 650 - 1000kbps so the space is still a concern as is compatibility with mobile devices. But you must be careful what encoder you are using and its settings if you want to test your ears. They are not all the same quality. For example iTunes can (if selected) filter out 10Hz and there is also the "Sound Check" for playback which is basically a volume leveling setting. There is also a thing called "Sound Encoding Adjustments".

Also my receiver had this funky setting for restoring "lost data" in compressed music. I leave it disabled as it sounds like night and day.

[audacity]

My main point is that we know (through science) that humans perceive a difference or show a preference to certain brands, or because they wish to support past decisions that they have made. I don't think any educated person would debate this, and this is one of the fundamental reasons why blind or double-blind experiments are done.

So, based on this we know that any A/B comparisons that aren't blind aren't very useful. I'm merely pointing out that there is a tool that you can use to "prove" to yourself that you can actually tell a difference because I'm betting that 99.9% of the people who believe they can hear a difference between lossy and lossless compression haven't actually done such a test.

I'm also aware that people have different capabilities in their hearing. For instance older people tend to not be able to hear as well as younger people. People can get hearing damage from listing to loud music. Sometimes people have too much wax in their ear to be able to hear certain frequencies. And, yes, different systems can produce different results. I generally recommend headphones for this type of testing because you can get higher quality sound reproduction for a lot less money than traditional loudspeakers.

Also my receiver had this funky setting for restoring "lost data" in compressed music.

How is this possible? Where does it get the missing data from if it isn't in the compressed file? Plus, isn't the audio usually decoded prior to getting to the receiver? I'd bet that this is just some sort of EQ effect/distortion that is being applied to the audio at the receiver. In other words, snake oil.

But if they have indeed figured out how to retrieve data lost that was lost in compression, perhaps they should go pick up their Turing Award.

[JamesK]

^^^^
This is why the "golden ear" types cannot be relied upon for accurate comparisons. They often make claims that are physically impossible to justify. A properly conducted A/B test is the only way to make a valid comparison, unless the differences are significant. For example, I don't need an A/B test to tell that the co-worker's radio is producing a lot of noise. It's blatant.

Not all hearing is equal, just as not all eyesight is equal

In the example of that Beatles song, your hearing would have to be fairly bad to not notice that difference I mentioned.

[dmoes]

Ok I think that there are so many factors here. when listening to most popular music I am in agreement that MP3s recorded at a reasonable bit rate say 192kbs that one would have a hard time hearing the difference. and in the majority of listeners very minor differences even with lower bit rates would not be disturbing or noticed without side by side comparison.

There are a few times that there would be a big difference. I recently archived a large collection of sound effects using at 250Kbps thinking this would be fine. most effects were fine. where I had problems was where most, was when the audio had steady sounds at a high pitch. example would be the hissing of steam cymbals. jet engines. they all sounded sort of watery and washed out. even at 320 they seemed off.

I ended up saving the whole thing as wav on a separate portable drive.

just my thoughts

[Jake]

How is this possible? Where does it get the missing data from if it isn't in the compressed file? Plus, isn't the audio usually decoded prior to getting to the receiver?

That was my point, it can't. It may be just looking for certain signatures typical of highly compressed audio and trying to mask the artifacts. Receivers have been able to directly read lossy bitstreamed data for a long time now. That includes WMA, MP3 and now AAC via a USB port or network interface.

Personally, I can't hear a difference and I suspect I would drop the bitrate to 192kbps if I was really pressed for space. Which I am not.

[Michael R.]

For the average song that I don't know particularly well, I can't tell if I'm listening to lossless or lossy. When we decided to copy our CD collection to a hard drive, however, we did some testing first. We have an expensive amp/speaker setup, so we felt it was worth the effort to know. We recorded a couple of songs in three formats: MP3, Apple Lossless, and the full original wave file. The difference between MP3 and the other two was obvious. The difference between Apple Lossless and the full wave file was negligible. We decided to copy our CD collection in Apple Lossless format.

We actually have two copies; Apple Lossless for listening at home and MP3/AAC for listening "on the go".

[57]

The difference between Apple Lossless and the full wave file was negligible

It shouldn't be negligible. It should be zero. Lossless is lossless.