How nice for the 1% who can drop $30K on music systems.
1 Like
1 Like
BB has sure enough blogged about the high-end stuff, whatâs so wrong about providing an honest counterpoint? As mutants, donât we all have different ears, different brains? Canât say that elf doesnât have a good ear for sound when heâs made this and this.
3 Likes
I did the same thing when recording my vinyl and tapes to CD. Even sped it up to see how much I could compress a too long concert tape to fit on a CD. I settled on 256 mp3 as decent considering how I listen to mp3. Also I didnât notice a 5% speed up.
ps Iâm not going to bother with this test since Iâd just listen to them on my old Dell laptop which sounds pretty awful most of the time.
1 Like
If you canât tell the difference, these Monster cables Iâm selling will be just the thing for you!
2 Likes
The thing is, Iâve never noticed much difference between my MP3s, and listening to bands live, so Iâm pretty dubious about the whole âIt brings out parts of the music Iâd never heard beforeâ. How does that square with removing any effect of the equipment on the sound?
Reminder that the results anyone personally gets from this test are meaningless as any individual result could be random chance. Only the distribution of everybodyâs answers could tell you anything.
1 Like
That was interesting. I got the first two wrong, picked the lowest quality then the 320 but repeatedly listened to them again and again and got the next 4 right no problem.
Something to do with the quality of the the background fill made the wav easier to pick out and by the last one I was in no doubt instantly.
Fun. My ears learned a thing.
FWIW I picked the WAV on the two tracks with naturalistic vocals (Neil Young and Suzanne Vega), and the 320k MP3 for the rest. This was on my monitorâs built-in speakers so the surprise is that I didnât pick any 128k MP3s (at 11:1 odds).
Obviously, however accurate it is, lossy compression embodies some subjective production choices, and some olden-tymes recording smiths mightâve done things differently if theyâd known about those choices. Even if the results were meaningful, theyâd only show that music engineered for MP3 sounds better than music that wasnât. Which, meh, that is the nature of progress, and yes, grandpa, having 250 hours of music in my pocket is progress.
Having said that, itâs funny to me that audiophiles should get defensive about that quip that they listen to the equipment rather than the music. Because it actually makes perfect sense to me that being hyper-focused on the reproduction would give you a deeper experience of the music. Itâs like, if you take a car apart and put it back together, it doesnât make much difference to how the car drives, but itâll transform the way you relate to it.
I correctly picked 1, 4, 5 and 6 on a laptop with somewhat above average internal speakers (as a non-audiophile, they seem to compete with my ultra budget Sennheiser buds), but only think that 1 and 5 had an appreciable difference, and even then it was only carefully listening with a direct comparison that I picked up on it. 4 was an educated guess, and 6 I donât think I would have gotten if it werenât for the fact that my bad internet connection made the larger WAV take a load of time to start. I kind of feel like the big difference for 1 and 5 was the crispness of the subtle details, that werenât getting smeared together quite as much. I also feel like 3 (and possibly others, canât be bothered to check) might not suffer as badly from compression because they use a lot of synthesised sounds, which intuitively seem like theyâd be easier to compress.
I convert all my CDs to FLAC and then high quality VBR MP3 for everything that has minimal space or that canât play FLAC, or where background noise renders any difference moot.
Itâs very hard to tell them apart, but itâs possible. MP3 definitely discards some of the lower bass content and itâs not entirely convincing with cymbals and other complex high frequency content.
2 Likes
Beyond the compression schemes and uncompressed WAV files we also have to consider the rig the audio was converted on as well as the playback equipment.
Sound cards have different signal to noise ratios (SNR) for playback and recording and they arenât all made equal. The total harmonic distortion (THD) of your recording and playback rigs is also an important factor. They also will support different sampling rates and playback resolution.
Even among higher end consumer grade playback equipment the range of SNR is between 100db and 124db which is considerable. Output THD may range from 0.002% to 0.0003% thatâs a huge difference. Frequency response on higher end cards can range from 10Hz to 90kHz @ 192kHz to 20Hz to 20kHz @ 48kHz. The numbers for low end sound cards and on-board are so bad they arenât worth mentioning.
So, even if you have good ears and decent headphones or speakers you may not hear the difference between the formats on your average sound card - especially those low end on-board audio outputs most people use.
Using $10 drugstore headphones I got 2/5 right. I would have gotten a third right because I heard a clear difference in the right choice but I wasnât entirely sure if the difference was actually better or just different. Other two I picked 320k. The lowest quality one was always very obvious.
In the case of Tomâs Diner you could say that MP3 was engineered for the music. In kind of an inside joke, the Suzanne Vega track is offered for testing in homage to the actual development of MP3 â Karlheinz Brandenburg thought he had the psychoacoustics nailed for symphonic material, but felt that a capella work would stumbleâŚthousands of replays later, including bringing Vega into the lab/studio, he got it. (And hence, you hear Vega hailed as the âMother of MP3.â)
3 Likes
I donât see it that way at all. Itâs treated as gospel among many music listeners that MP3s are âbadâ and sound poor compared with lossless formats. See for example (as the article notes), TIDAL offering lossless streaming for twice as much money. I guarantee that many if not most of the people paying that are doing so under the belief that itâs âbetter,â without necessarily having the high-end equipment that would make the difference that apparent.
The point of scoring 320 MP3 as a loss is to show that, no, on that sample at least you could not distinguish high-quality MP3 from lossless, no matter what youâve been told. That was the whole point of the test.
[quote=âSamSam, post:37, topic:58935, full:trueâ]
The point of scoring 320 MP3 as a loss is to show that, no, on that sample at least you could not distinguish high-quality MP3 from lossless, no matter what youâve been told. [/quote]
I stand corrected! : )
The link I had followed earlier that day indicated the point of the article was to show that 128Kbps was insufficient. Re-reading the title, I agree with your assessment
I got 4/6 right, but I would call it luck. It was very difficult. Even having just finished mastering a CD project, I didnât expect to do better either.
What I donât hear many people talking about is that most modern recordings are done on 96KHz/48bit or higher (192KHz) digital converters, as was our project. The digital mix sounds amazing, the detail is really stunning. In our case, we mixed the digital mix to an Ampex 1" tape, which was also amazing, and mastered from there.
However, then we make a HUGE down sample to CD, which is 16bit 44KHz. This IMHO is where the big loss occurs, not going from CD to 320K or 192K mp3. 128k mp3 is another issue though, it is just not enough bandwidth to represent complex music, and I think most people can tell that difference.
Most mortals will never have the opportunity to hear the digital (or tape) masters. Distribution is done at CD quality. Older masters just arenât available, nobody wants to dig them up and re-distribute them. New projects are down sampled to CD (Redbook audio). If you want to hear the digital mix, you pretty much have to be sitting in the studio that mixed or mastered it - they really donât leave the premises.
Again this is just my opinion, but a well processed mp3 is not much different than Redbook audio, so this whole Tidal high quality audio is not a business, and if it is, it is a snake oil business.
IF (and itâs a big IF) the digital mixes, straight from the mastering DAW converters or a 192/48 ADC of a master tape were available, that to me is worth something. I am pretty sure most of those who struggled with the NPR test could hear the difference - it is pretty stunning.
3 Likes
Much more is lost due to mastering than from downsampling. Testing that inserted 16/44.1 segments into 24/192 proved reliably inaudible to all listeners. 16/44.1 is, for all intents and purposes, really all you need for transparent audio reproduction.
Yeah, anything over 16/44.1 is a waste of time in terms of audio playback.
16-bit (with dithering) allows for the reproduction of the maximum dynamic range the human ear can deal with (120db, which is loud enough to cause permanent hearing damage pretty quickly, around the volume of an airplane taking off). 24-bit would allow you to remove the dithering step maybe (because the distortion the dithering normally removes would be below the noise floor of the electronics of any system you would be using), but you can just dither, so why bother? Even without dithering 16-bit gives you 96db, which is well above the dynamic range of vinyl or tape, and probably more than enough for most purposes.
44.1 can capture the highest frequencies the human ear can hear, in theory, so thereâs not much point in going above this. In practice though AD converters will sample at very high frequencies (to avoid using analogue filters which roll-off in the human hearing range, which if theyâre not expensive high quality ones will add noticeable distortion, you can then use cheaper digital filters instead); these ADCs will immediately down-sample to the requested rate.
Higher bit rate and sampling rates have there uses at various stages of the audio chain, from recording to digital production environments (where extra dynamic range is needed to sum large numbers of signals and avoid clipping, and bit depths of 32/64 are more convenient to work with anyway), extra frequency data can also be handy to have effects processing. Once it comes to mastering the final output though, thereâs no need for anything greater than 16/44.1.
3 Likes
I wasnât in the mood to type all this out, so thanks.
