Piano World Home Page Forums Our Most Popular Forums Digital Pianos - Electronic Pianos - Synths & Keyboards OT audio quality, can you hear the difference?

I've been seeing a lot of Qobuz ads on my Facebook feed. I have a paid subscription to Spotify and am happy with their service and the size of their library.

I took the following test and scored three out of six. At the very least, I nailed the piano selection though without the comparisons, I would have been happy to listen to the ... inferior format.

https://www.npr.org/sections/therecord/2015/06/02/411473508/how-well-can-you-hear-audio-quality

I've done a lot of tests like that and my result is always at around 50%. Just for clarification, in tests like that a result of 50% means that you don't pass the test at all since even if you clicked randomly you would still get around 50% because that's how probability works One should consistently produce high scores like more than 80 or even 90% correct guesses to consider the test as passed.

I don't see how you would score this test.
That web page asks: "For each song, listen to all three samples and choose the one that you think is the highest-quality audio."

I got 100% on this.
When asked for my opinion how can I possibly be wrong?
I cannot be wrong! I'm 100% correct. Always.

So, if you choose one in three files, then the random-click chance would be 33%, so 50% is not a bad result.

The only selection that I might have bet money on was the piano selection. There was a very slight difference that I could hear (or thought I could hear). I would imagine small classical groups like a string quartet would also highlight subtle differences in the various audio formats.

I can also imagine if they cheated and used the same exact file for all three examples ... and if one of those examples were 1/3 dB louder, we'd choose the louder of the three even though they were identical in all other respects.

50% on the presented test is just 3 correct answers, just 1 more than the most probable scores when answering randomly. I won’t consider it a good result.

However 50% in a similar test with let’s say 20 questions would be more a sign of a heard degradation of MP3.

Now you sound like my missus . . . .

Distinguishing 320kbit from Flac or wav is VERY difficult and only really works if problem samples are present.
You have to listen intently to the background and crank the volume way up. I managed to hear it once but not in this test.

During casual listening at my typical low volumes, everything above 192kbits sounds the same to me (barring problematic samples). I can hear compression artifacts when listening to YouTube.

In the test of the OP I listened to the piano piece only and was not able to hear any difference between all 3.

No, really. How can I be wrong when asked which sound I think is best? I won't lie. I'll give an answer. So what's the point of scoring?
You, too, will get 100% on the test, because you'll say which sound you think is best.

Seems like 320 kbps is enough for me. With a quick listen, I chose it 4 times out of six. Once I got the uncompressed and with JayZ I thought the 128 kbps one sounded best (maybe it's the warbling background that's kinda like compression . But with only 1/6 being fooled by the lowest rated, I'd say there it at least a bit of a difference.

Mac is being somewhat pedantic about the poorly phrased question



Anyway I picked 4 uncompressed, 1 320kbps (the coldplay) and 1 128kbps. Given that was the one I was given first (it randomises the order) and was the Suzanne Vega track that was one of the key bits MP3 was tested on I'll let myself off that.

I did have my headphones up considerably louder than I normally would and most of the picks between the 320 and the uncompressed were far from certain.

I'm curious what encoder they used as that makes a huge difference, I'd expect with a vaguely recent version of something like LAME that it should be a struggle to tell at 192, let alone 320 so my suspicion is that they didn't use something that good.

5/6 ---- Although I suspect the less then WAV files may have been tweaked with some EQ.

I still think the "Star Wars" test is the best demo to someone who doubts Vinyl vs CD, etc. I discovered this in the age of VHS, CD and Laserdisk... hmm, interesting. The biggest fault is usually the ambiance and the loudness losses its peak from the original. But CD/MP3 is a tried and tested system.... uh ha, $15USD please.

Yes, I saw this test a while ago too..
The great thing about this test is, it randomizes the file orders every time, so nobody can memorize or anticipate the correct answers. It's really a blind test every time and forces you to listen.
First time I tried it playing through laptop speakers it was hopeless...
Switching to my studio headphones and outboard DAC and turning up the volume, I could just about hear the differences.
I found the 128kbps was usually easiest to spot and rule out, but the other two sounded quite similar until you really concentrate on the transients and the extreme HF.
After quite a few test attempts, my score gradually improved, until finally I got the hang of it and could hear 6 out of 6 correctly every time I did the test.
The WAV is the only one that fully extends up into 20kHz range, whereas the MP3s are more limited bandwidth except on very strong transients. 128kbps averages about 15kHz top end cut off, but the 320kbps is probably about 18kHz cut off.

My ears are apparently broken. The only one I could tell was any better was the classical selection, and I got it right multiple times after reading the comment about how they change positions. All the others I seemed to prefer the 128k sample by quite a large difference. Not once did I pick the uncompressed for anything other than the classical.

I don't think the test is very helpful as the music production quality varies so much among the songs.

The Mozart Concerto & Suzane Vega bits were well recorded and mastered so quality differences to mp3 should be more obvious and were to me.

The JayZ, Katy Perry and Coldplay recordings were terrible as "uncompressed WAV" files. Super-processed, super-compressed, low dynamic range music always sounds flat and lifeless. And sounds equally terrible when compressed. I owned a few Coldplay CDs and was always irritated by the poor production work.

If one listens to concertos, do the test only with concertos. If one likes jazz, do the test only with jazz. etc.

I only tried two tracks (Murray Perahia and Coldplay in that session), but managed to select the Uncompressed WAV on both occasions. Maybe I was just lucky. I noticed that in the Coldplay tune, there is narrowing of the stereo image at different points of the song - a detail that seemed to be lost by the MP3 encoding.

Kind regards,
James
x

Well my hearing stops at around 12.5KHz. Maybe that's why it is harder for me to hear.

Tests like this don't tell us very much, because there are so many unconstrained variables. MP3 compression, at least, has been tested in lab conditions for at least ten years, and the consensus seems to be that for most people, with most material, the cut-off point where it becomes impossible to tell the compressed from the uncompressed material is somewhere between 128 and 192 kbits/sec. The question of which question a person "prefers" is even more complicated.

This figure of 128-192 kbits/sec is based on the use of lab-grade equipment in closely controlled listening conditions. I suspect that it's even harder to distinguish compressed from uncompressed audio with consumer apparatus.

Of course, these are typical figures, and I know of one person who can reliably distinguish 192 from 320 kbits/sec MP3 recordings when played on an ordinary domestic hi-fi. Unless he's cheating somehow, e.g., by reading my mind I suspect that this is something you could train yourself to do, if you really wanted to, and if your high-frequency hearing range extends far enough (you'd probably have to be under 20).

There are many great tests at this website:
http://abx.digitalfeed.net

Click on the other tests. For instance the one for MP3. Using my best equipment: Apogee Groove DAC/Amp and Sennheiser HD-650 I was consistently able to guess with almost 100% certainty only MP3 at 128kbps compared to WAV. Starting with 160kbps I was able to hear difference on only one track while the rest are indistinguishable to my ears. I believe 192kbps are a nice starting point. Better even would be a VBR that varies bitrates depending on complexity.

Now, I know an aidiofool who claims to hear difference between FLAC and WAV Schizos around us everywhere

2/6 here.
The first sample has an evident hiss in the wav file, that made me click on the best mp3 as I thought it was an artifact.
I selected best mp3 over worst in all but one ocassion.
I nailed the piano one

With cheap in-ear headphone and an office pc you can't do better, I'm afraid...or it's just me. Who cares.