Do you really need those danceable cables?
June 3, 2015 11:22 AM   Subscribe

Now that Tidal has given us a CD quality streaming service, NPR Music Editor Jacob Ganz and some co-workers have put together a quiz to help figure out if it's worth paying for: How Well Can You Hear Audio Quality? (Tidal has its own test, if you want something different.) "Danceable cables" come from an old Gizmodo review.
posted by Going To Maine (69 comments total) 15 users marked this as a favorite
 
Amazingly to me I consistently picked the 380 as sounding the highest quality to me. 5/6 in fact.
posted by Carillon at 11:27 AM on June 3, 2015


The first thing I heard about Tidal was Jay Z being upset it wasn't getting good press reviews and he implied it was because the record companies and Apple paid the tech bloggers to write negative ink. I wish I could find the article because it was quite funny. The poor tech blogger was wondering where his check was.
posted by cjorgensen at 11:27 AM on June 3, 2015 [1 favorite]


5/6 here, but I'm going to attribute that to years of being around recording studios and mixing boards. I know how to listen for these things, but that doesn't make me more appreciative of the actual recording or performance. Rather, it makes me appreciate the mixing and engineering that took place afterwards.

I kind of don't like how this forces listeners to think like engineers... Things along the lines of... "The cymbals are tinny sounding, therefore this level of compression is 'inferior'"...

Does that make the art better? No.
posted by raihan_ at 11:30 AM on June 3, 2015 [3 favorites]


Great that they included "Tom's Diner", a nod to that track's being used as a benchmark in the early development of the MP3 format.
posted by RobotVoodooPower at 11:32 AM on June 3, 2015 [4 favorites]


it's a pity that the .wav file always buffers on this connection; that spoils the test for me
posted by thelonius at 11:35 AM on June 3, 2015 [4 favorites]


Listening on Apple earbuds, at work, with office noise all over, I picked either the 320kpbs or the WAV about equally (honestly, they're indistinguishable unless you've got something better than earbuds to listen on), but I picked the 128kpbs for the Coldplay clip because they all sounded so bad (pretty sure I wouldn't have made this error on my normal setup at home).

I just don't think 320kbps MP3 and uncompressed audio sound different enough--on most sources--to justify the extra bandwidth / disk.
posted by uncleozzy at 11:36 AM on June 3, 2015 [3 favorites]


Got 5/6 when I took the test yesterday and was quite surprised. I've spent a lot of time listening to podcasts and audiobooks through some cheap Sony earbuds and I assumed that hadn't done my ears a lot of good.
posted by fuse theorem at 11:36 AM on June 3, 2015


I was bamboozled, it's true! I have good headphones and a good DAC, and all my selections were essentially random.
posted by JoeBlubaugh at 11:38 AM on June 3, 2015


0/6. I'm 43 though. Most times I picked the lowest quality one.
posted by colie at 11:39 AM on June 3, 2015 [3 favorites]


I got 4 out of 6 on my MBA, so I guess I have super ears even with the tinnitus. Suck it, plebs.
posted by entropicamericana at 11:39 AM on June 3, 2015 [1 favorite]


On the other hand on Tidal I got:

."One in five correct, whoops!
Are you sure you have connected your audio system correctly? Check and try again.
TEST AGAIN"

Thanks for the condescension!!
posted by Carillon at 11:42 AM on June 3, 2015 [3 favorites]


5/6 (with the one wrong being a 320) even on computer speakers, which NPR says means I'm either extraordinary or lucky. As I've been playing music (non-professionally, but at times at fairly high levels) for 20 years, and now essentially make my living with my ears, I'm hoping its the former, or I'm in trouble.
posted by damayanti at 11:43 AM on June 3, 2015


I've always been a "320 kbps should be enough for anyone" partisan.

Then I saw this demonstration, which involves a playing a 320 kbps mp3 against a phase-inverted uncompressed wav of the same track, such that you can only hear the difference between the files.

It's—well, listen for yourself. If nothing else, it's a fascination demonstration of what you can reveal in audio via phase-inversion.

I'm too far invested in compressed audio for my personal listening library, but there's certainly a case to be made for consumer-facing lossless compression.
posted by Sokka shot first at 11:46 AM on June 3, 2015 [10 favorites]


4/6 through headphones plugged into my laptop. But it was much harder than I expected and the two I got wrong (sorry, Neil!) I picked the 128. Really interesting. I feel schooled.
posted by The Bellman at 11:50 AM on June 3, 2015


I don't care whether I or anybody else can tell the difference between anything, and I'm starting to get suspicious of how frequently these things are popping up in my various feeds. It always carries this implication, spoken or otherwise, that we don't really ever need to distribute audio above 320, or that lossy is always "good enough".

Why does it matter what is good enough, what is the pressure in the first place? The storage & bandwidth constraints that made lossy compression a necessity in the first place by and large no longer exist. It's trivial to stream uncompressed CD audio. I feel like this has nothing to do with quality, after all, that there's some other industrial reason to get people to accept that the basic digital audio standard established decades ago is just something for snobs that us regular folk don't need.

I don't understand in a world of instant HD video, music is still being subjected to this kind of nickle-and-diming. What people want or think they can discern is irrelevant. In any other medium, you just give people what is technologically possible, and then they accept it as the new normal and don't want anything else.
posted by anazgnos at 11:50 AM on June 3, 2015 [18 favorites]


I read about this test on another site earlier; pre-listening prediction: I will be able to distinguish between 128K and 320K, but probably not between 320K and uncompressed (i.e., I'll get 320 confused with WAV most times).

Actual results: basically that. I picked the uncompressed version as "best" 4 times out of 6, and the other two times I picked the 320K MP3. This is consistent with my experience / preferences when ripping my CD collection however many years ago. Listening with Bang & Olufsen laptop speakers - may try later with headphones or a full system with sub.

I wish they wouldn't give green checks and red X's for "right" and "wrong." I feel like a better way to frame it is what level of compression can you, personally, detect, rather than whether you are able to pick the "best" of the 3. If you can't distinguish, then obviously the uncompressed stream is not "best" for you in any relevant sense.

And:
it's a pity that the .wav file always buffers on this connection; that spoils the test for me

Indeed. Only happened once for me, but was a dead giveaway.
posted by Joey Buttafoucault at 11:52 AM on June 3, 2015


Sokka shot first: I can't see where he states what the bitrate of the MP3 file is. Am I missing it? Without that crucial information that test is completely useless.
posted by leo_r at 11:55 AM on June 3, 2015


It's trivial to stream uncompressed CD audio

... doesn't really jibe with:

it's a pity that the .wav file always buffers on this connection

Also, I'm not sure that we really live in a world of instant HD video, in particular in the US. Bandwidth to the home is fairly limited, and the 6Mbps video Netflix is pushing doesn't fit down most of those pipes. Most people are getting, on the best days, the 320kbps MP3 version of the HD video.

Point being, good-enough is good-enough because bandwidth is still finite, and most people would rather not wait for their uncompressed audio to buffer when they're listening through earbuds on the subway or in the car.
posted by uncleozzy at 11:56 AM on June 3, 2015 [7 favorites]


I don't understand in a world of instant HD video, music is still being subjected to this kind of nickle-and-diming.

Well, for one thing we don't all have unlimited 4G data on our phones (hell, we don't all have 4G all the time depending on our provider godihateyouTmobile). I personally appreciate high-quality compressed audio when I'm out and about so I can listen without really worrying about whether I'm going to hit my monthly data limit.

On a fixed connection though, yeah, I'd also prefer uncompressed versions.
posted by Joey Buttafoucault at 11:56 AM on June 3, 2015


I got 0/6 using AmazonBasics earbuds on my work computer with whatever generic onboard audio it has. I'm 43 and I can't quite hear 17khz test tones on this setup.

I am an electronic musician. I was listening for telltale signs and I wasn't finding them.

(I have noticed these earbuds are not as good as the Sony cans I used to have here. Given the excessive warmth here at times I couldn't handle big foam pads covering half my face anymore, and I don't know where my Sennheiser earbuds wandered off to.)

At any rate, I suspect under ideal conditions I would still not hear enough difference to worry about it with most material. Certainly not if I was listening to it and appreciating it as music instead of trying to prove either that my ears are good or MP3 is bad.


laying a 320 kbps mp3 against a phase-inverted uncompressed wav of the same track

I've seen similar demonstrations, and they always have to crank it up to point out the difference, and then they treat that difference as if it's shocking for some reason. MP3 compression is supposed to remove information that, due to psychoacoustics, you won't notice is gone; it does not guarantee an identical signal. So this is just proving that it works, minus the part where you get to hear A/B comparisons.
posted by Foosnark at 11:57 AM on June 3, 2015 [12 favorites]


Got 4/6 on the NPR test and 4/5 on the Tidal test. My work headphones are cheap Sony loopy ones (but they still have great bass sounds), and I require an audio jack extension plugged to the tower. But I still had to replay multiple times for each quality to determine the best one. Overall I agree that it really doesn't matter if you don't have a set up designed to amplify the differences.
posted by numaner at 11:57 AM on June 3, 2015


(And for what it's worth, I hated 4 of those 6 tracks anyway.)
posted by Foosnark at 11:58 AM on June 3, 2015 [2 favorites]


In any other medium, you just give people what is technologically possible, and then they accept it as the new normal and don't want anything else.

That's why Netflix streams out raw Blu-Ray video data for every movie, all 50GB of it. Wait, they don't?

I'm exaggerating, of course (not every Blu-Ray has 50GB of video data), but there are always limitations. For example, I've ripped all the music I've bought since 2013 to FLAC, but there's no way I could store my whole FLAC library on my phone. Streaming those FLAC files to my phone is a non-starter, not just because of bandwidth concerns (the biggest mobile plans are about 6GB a month, which means you can listen to about 20 albums' worth of music if you do nothing else with your phone online) but also due to signal strength issues, depending on how crappy your carrier is.

And though Netflix streams full HD to you, not all 1080p video is equal. Depending on the bitrate they've been encoded at, there can be wild variations in image quality even though you're getting the same number of pixels.
posted by chrominance at 12:00 PM on June 3, 2015 [3 favorites]


I'm disappointed that these aren't set up as ABX tests. These tests pretend to test whether you can hear the difference between high and low quality audio; what they really test is whether you can hear the difference and whether you know which is which.

What I'm saying is, I also think "high quality" Coldplay isn't.
posted by lozierj at 12:07 PM on June 3, 2015 [4 favorites]


Then I saw this demonstration , which involves a playing a 320 kbps mp3 against a phase-inverted uncompressed wav of the same track, such that you can only hear the difference between the files.
This sounds like it should be a useful demonstration of something, but it really doesn't tell you anything. That horrible noise you're hearing is technically "present" in both, and neither, of the versions. The difference between two things isn't a quality of either one on it's own.
If you invert one side of a stereo track and combine it with the other side, it will sound like crap (no bass, no vocals, too much reverb), but clearly this doesn't mean that one of the channels is bad.
posted by calmsea at 12:10 PM on June 3, 2015 [7 favorites]


Then I saw this demonstration , which involves a playing a 320 kbps mp3 against a phase-inverted uncompressed wav of the same track, such that you can only hear the difference between the files

It's a very interesting demonstration of coding artifacts, but you're not 'missing' that detail in the 320kbps file so it's not really a good demonstration of how crappy (or otherwise) MP3 or perceptual codecs in general are. All of the audio in the 'difference' track is what your auditory system masks out when listening to the full quality track. You literally* cannot hear those frequencies, because they are too close to much louder frequencies within the same critical band.

* "You" being a typical listener, ymmv

I would be pretty confident is saying that 95% of the population would not be able to tell the difference between a 320kbps MP3 and WAV in an ABX test in a good listening environment (unless it's a particularly difficult track to encode e.g. castanets is a classic example). At 128kbps it's a lot more noticeable, but newer, more advanced codecs at 128kbps would be almost as difficult to discern as the 320kbps MP3 compared to wav.

I got 3/6 both times I took the test (on headphones in a noisy and quiet environment), but I also selected 128kbps at least once in both attempts, so I doubt I would do much better than chance with a larger sample size.
posted by TwoWordReview at 12:12 PM on June 3, 2015


Also, I'm not sure that we really live in a world of instant HD video, in particular in the US. Bandwidth to the home is fairly limited, and the 6Mbps video Netflix is pushing doesn't fit down most of those pipes. Most people are getting, on the best days, the 320kbps MP3 version of the HD video.

Instant HD video is a really interesting example, as two --probably related-- corners are still being cut there, that I find at least as annoying as over-compressed music:

1) I find that whenever there's a lot of movement on the screen, like a chase through a leafy forest for instance, things still get pretty pixelated in a way they don't on BD or even DVD for the most part.
2) I have yet to find a device or streaming service that offers to build a bigger buffer on a "slower" connection (often theirs, not yours), rather than downgrade the image. I still get the occasional Amazon or Netflix show that will downgrade to VHS resolution, rather than use the memory (internal or add on through SD card) on my Roku to preload more of the HD version. I think "because piracy" is the excuse for this one.

Streaming video and/or the reliable bandwidth to support it still have a ways to go.
posted by Lentrohamsanin at 12:16 PM on June 3, 2015


Then I saw this demonstration , which involves a playing a 320 kbps mp3 against a phase-inverted uncompressed wav of the same track, such that you can only hear the difference between the files.

MP3 does not preserve precise timing or phase information, so I would expect results like this. But really doing mathematical analyses on the effects of mp3 are missing the point of mp3. Of course there are differences that you can detect mathematically, or hear in isolation. The whole point of the psychoacoustic model of compression is that those differences are things we are not able to perceive in the context of listening to the full encoded audio. Blind A/B testing is really the only effective way to evaluate this; any other analysis is just comparing one model of how we perceive sound to another.

The storage & bandwidth constraints that made lossy compression a necessity in the first place by and large no longer exist.

I don't agree with this. The amount of music I carry on my phone is for sure limited by storage. Maybe from my end, for streaming, I usually have some extra bandwidth available, but from the provider's side of things, well, YouTube's 480p videos often have a lower total bitrate than uncompressed CD quality audio.
posted by aubilenon at 12:17 PM on June 3, 2015 [4 favorites]


I have yet to find a device or streaming service that offers to build a bigger buffer on a "slower" connection (often theirs, not yours), rather than downgrade the image.

Most streaming video devices are relatively small and use nearly all their buffer space all the time. There's nowhere to put a bigger buffer.
posted by GuyZero at 12:24 PM on June 3, 2015


The advice I usually hear nowadays is to download and store FLAC or other lossless format, but to encode from the lossless files to a lossy format for portability. There's a big difference between the MP3 encoders from back in the day and AAC encoders today. It would be nice to be able to go back to the lossless well to encode to a more efficient format for better quality at the same bitrate to carry around on my phone, but I'm stuck with the MP3s I ripped 10-15 years ago.

Also for those who liked the sound of crappy low-bitrate encoding, well there's an app for that...
posted by TwoWordReview at 12:26 PM on June 3, 2015 [4 favorites]


6/6. And, to me, it took seconds on each track except for the Coldplay, which was so compressed that it wasn't until the cymbals kicked in that there was enough dynamics to really show the difference.

And, I'm not on an extreme high end system, but I'm not on crap either -- Modi 2U DAC, Magni 2U Amp, Grado SR325 cans, well beaten up ears.
posted by eriko at 12:36 PM on June 3, 2015


The advice I usually hear nowadays is to download and store FLAC or other lossless format, but to encode from the lossless files to a lossy format for portability.

Ding ding ding

320 sits in an uncomfortable valley for me - if I'm putting stuff on my ipod and space is limited, it's too big. If I'm at home and space is relaxed, it's unnecessarily lower quality than the master and has the handicap of not being exportable to any other subsidiary format.

I'm just generally distrustful that what is good today is good forever in lossy compression. In 1996, a 96kb/s mp3 sounded pretty amazing. I'd never heard anything like that from a mere downloadable file before, and it only took an hour and a half! I had no idea what an artifact was. It takes time for those things to jump out at you, and without fail they always do. I might have had things that I listened to 20 times before the phasing or swirlies started to poke out, and once they did they were impossible to ignore. It's been a hard-learned lesson to always have the master to go back to.

What I really question is just the presentation, or the narrative. We don't see apple going "hey check out our new camera - it captures less colors that the old one, but you won't even know the difference!"
posted by anazgnos at 12:38 PM on June 3, 2015 [1 favorite]


calmsea: "If you invert one side of a stereo track and combine it with the other side, it will sound like crap ..."

Except for Brian Eno's Ambient 4: On Land which was mixed explicitly for the stereo inverted and combined signal to be played on a third speaker as a kind of poor person's quadrophonic. Yes, Eno is weird.
posted by thatwhichfalls at 12:52 PM on June 3, 2015 [5 favorites]


I have some 96kHz/24 bit flacs of vinyl rips around here somewhere...
posted by mikelieman at 1:05 PM on June 3, 2015


Amazingly to me I consistently picked the 380 as sounding the highest quality to me. 5/6 in fact.

Me too. I didn't get the WAV once. I did pick 128 for Tom's Diner which I guess is a good sign about this nifty new MP3 thing we're working on! I didn't expect to be able to tell the difference between 320 and WAV anyway but I did expect to be able to distinguish 128 more easily - 6 tracks isn't very many though and I wasn't especially careful about it and also this:

I'm disappointed that these aren't set up as ABX tests. These tests pretend to test whether you can hear the difference between high and low quality audio; what they really test is whether you can hear the difference and whether you know which is which.


A few more thoughts:

- I'm in my mid-20s and I've probably listened to many, many more MP3s in my life than CDs. So that probably affects the way I expect things to sound.

- Expectation matters a lot when you're trying to pick "best" and "worst." I make electronic music and there's a track on my old Soundcloud (which uses 128K MP3) that I came to dislike only a short while after uploading it. How could I have thought that was good enough to share? Except then later I listened to the original WAV on my laptop and it's still the track I thought it was - that white noise buildup is clear and crisp instead of murky and flangey, the hi-hats don't sound like shit, the subtle distorted high-end buzz from the bass synth is actually there at just the right level. But this makes me wonder if maybe that stuff doesn't matter to people who don't know how it's meant to sound? Or does it say that you shouldn't count on people being able to hear stuff over 8K the way you intend it anyway?

- Or looking at expectation another way, from that experience of working with audio a lot - but not in a recording capacity - I'm used to hearing acapellas that are messily ripped from god knows where god knows how, older studio tracks with bleed from the band, or remix contest giveaways fully drenched in effects. So something like Tom's Diner that's supposed to be a super clean sparse isolated vocal? What is that like even?

- I agree that this selection of music (the NPR) isn't a great choice to show the difference. I'd guess on the other hand the Tidal one (which I didn't do) probably tries very hard to show the difference so maybe the NPR one is making a point here. Give me a ride cymbal and the 128 tends to become clear.
posted by atoxyl at 1:09 PM on June 3, 2015 [1 favorite]


Ah, audio quality tests, the Pepsi Challenge of the new generation.

I suck at these things for a reason: I need some familiarity with the songs (at any quality) to notice things. For instance, I didn't notice I had a few albums in 128kbs until I started to listen to them more and wondering why they sounded so flat. My own stuff, I can usually tell quickly if the quality sounds off, then getting fits when it is off.

I have no need for ultra-high quality audio files, but I'm not an archivist, either. And if I was, the question would be "how can we improve the current state of lossless audio", not exactly Tidal Challenges.
posted by lmfsilva at 1:14 PM on June 3, 2015


5/6 on some Sennheiser earbuds - I did see the wav buffer once but it was pretty clear when comparing side by side.
posted by zeoslap at 1:15 PM on June 3, 2015


2/6. It's comforting to know that I don't have to worry about audio quality as I can't discern it.
posted by 23skidoo at 1:21 PM on June 3, 2015 [1 favorite]


2/6 right, I ended up picking 2 WAV, 2 320k, and 2 128k tracks as the best.
It would appear that I'm a music egalitarian!
posted by CrystalDave at 1:41 PM on June 3, 2015


Far more scientific double-blind tests on sites like HydrogenAudio have shown that even on good equipment, most people can't distinguish between 128kbps VBR MP3 and lossless. Not sure why people are getting good results here.
posted by archagon at 1:54 PM on June 3, 2015 [1 favorite]


> Then I saw this demonstration yt , which involves a playing a 320 kbps mp3 against a phase-inverted uncompressed wav of the same track, such that you can only hear the difference between the files.

I've seen this before, but the explanation I've heard is that the differences are almost certainly inaudible when the rest of the sound is added back in, given the limits of the human ear. MP3 is designed that way.

EDIT: Or, what everyone else said.
posted by archagon at 1:56 PM on June 3, 2015 [1 favorite]


I got 1/6 right, favoring the 320kbps MP3 in four out of the five I got wrong (I very easily got Suzanne Vega right, and I essentially punted on the Katy Perry because it was so fatiguing a mix I couldn't be bothered to try hard). My comment there has sort of vanished due to age, but I do wonder what would happen if they had been able to use original analog sources instead of CDs, and down convert from higher resolution files. Almost everything they chose was a poor example of digital encoding in the first place (the Mozart is an early digital conversion, before anybody knew how to get the best results from the tools; three of the recent tracks are extremely compressed due to the loudness wars), so I'm less lamenting my own tin ear and more wondering if there's not a way to come up with a better test. Telling me I not only can't tell the difference between a WAV and a 320Kbps MP3 but prefer the MP3 isn't so useful, when the recording itself is rather poor.

I would rather know if I can tell the difference between, say, 192/48 and 44.1/16, if all other variables are equal (including loudness). But that's not what they're testing here.
posted by fedward at 2:02 PM on June 3, 2015


I have yet to find a device or streaming service that offers to build a bigger buffer on a "slower" connection (often theirs, not yours), rather than downgrade the image

I think the reason they limit prebuffering is to both save their own bandwidth needs, and to flatten spikes in it when people with fast connections watch things. It saves bandwidth because it's pretty common for people to stop watching before the end, and if the service's bandwidth is the bottleneck as you say, then doing more prebuffering will just degrade their service for everyone.
posted by aubilenon at 2:04 PM on June 3, 2015


I honestly couldn't tell between the tracks in the NPR one, I didn't even bother to guess. I got 4/5 on the tidal one, though, but I suspect it was luck because if any difference was there it was minuscule.
posted by Pyry at 2:06 PM on June 3, 2015


Got 4/6 yesterday on the NPR test out of my computer speakers. I blame the two losses on the fact that Jay Z and Katy Perry are so brickwalled it's hard to determine any fidelity in them. On the classical, Neil Young, etc the difference was a lot easier to catch.
posted by Ber at 2:13 PM on June 3, 2015


I figured it'd be easy to tell on my iPad with Grado SR80 phones, but it wasn't. I think that part of the takeaway is that differnces in sound quality don't always manifest themselves across all sorts of tracks and music. On preview what atoxyl said.
posted by persona au gratin at 2:17 PM on June 3, 2015


The WAV files aren't loading for some reason, so I tried to see if I could hear the difference between the 320 and the 128. I can't. I consistently picked the 128 as the higher quality audio. Not sure what this says about me or my headphones, although I will note that before moving to NYC, I had some of the best hearing of people I knew (in terms of hearing higher pitches). Not sure what 6 years of subway trains have done, but there's that.
posted by Hactar at 2:17 PM on June 3, 2015


WHAT?!
posted by I-baLL at 2:23 PM on June 3, 2015 [1 favorite]


The Neil Young and the Katy Perry were obvious to me. I'm not sure why, but on those two tracks the uncompressed file sounded markedly better than the other two.
The Coldplay was tougher because all three sounded distorted and sibilant on the "s" sounds. I picked the high bitrate MP3 but it was almost a guess. Is that just a poorly recorded album?
Tom's Diner should have been easier but it wasn't. They all sounded pretty good.

A better test would have included a live performance with audience applause, which is almost impossible to compress without horrible sounding artifacts.
posted by rocket88 at 2:34 PM on June 3, 2015


Then I saw this demonstration , which involves a playing a 320 kbps mp3 against a phase-inverted uncompressed wav of the same track, such that you can only hear the difference between the files.


Not a good test at all.
Take a WAV file and time shift it by 23 microseconds (one sample period). The resulting shifted file is identical to the original in every way but the time shift. If you then phase-invert one and add it to the other, you will hear artifacts similar to those in the demonstration, yet nothing was added or removed in the modified file.
posted by rocket88 at 2:41 PM on June 3, 2015


It's crazy to me that people are saying it's obvious. I'm clearly fairly untrained but I have no idea how it can be obvious. Wish I had that talent.
posted by Carillon at 2:44 PM on June 3, 2015


And if you're tempted to use your browser's inspector to find URLs for each sample, download them, plug them into Audacity, and then compare the waveforms, perhaps performing just that inversion test … just note that the MP3 files don't have exactly the same amount of silence the WAV files do, so your first order of business will be to trim them precisely (Ask Me How I Know!).

Good luck!
posted by fedward at 2:47 PM on June 3, 2015


Do you really want to be able to tell that a song sounds bad?

Why do people spend so much money on audiophile equipment just to scoff at their own files?
posted by oceanjesse at 3:15 PM on June 3, 2015


And, I'm not on an extreme high end system, but I'm not on crap either -- Modi 2U DAC, Magni 2U Amp, Grado SR325 cans

Dude, I say this with nothing but love and budget-audiophile camaraderie, but don't get it twisted: you have a nicer music-listening setup than, statistically speaking, almost everyone in the world. You are an audio one-percenter.

It's true--some other guys have Koetsu cartridges or whatever. It is also true that millions of other people have Apple earbuds.
posted by box at 3:22 PM on June 3, 2015 [1 favorite]


First test, using Shure SRH-440 cans plugged into my work computer: 4/6. I think I picked the 320s for the Neil Young and the classical piece.

Second test, just now, using M-Audio AV30 speakers plugged into my home computer: 3/6. For Jay-Z and Coldplay, I picked the 128s, and for Suzanne Vega the 320.

I don't really know what this says about my ears.
posted by chrominance at 3:28 PM on June 3, 2015


Why do people spend so much money on audiophile equipment just to scoff at their own files?

It's a relentless pursuit of "the best," for whatever value of best applies (and that's not the same for everybody). Personally I only spend as much money as I have to to be able to listen to something without being irritated by it, which is different than the stereotypical audiophile approach of buying the best thing one can afford (and agonizing over whatever constitutes "best" in that case). My experience, though, is that as soon as one thing is good enough not to be the problem, the flaw with some other thing becomes more apparent. In my case, nice headphones made low-bit-rate MP3s obvious, but 256Kbps AAC usually sounded "good enough" even through the nicer cans.

On the other hand, the Loudness Wars have made a lot of recorded music completely unenjoyable for me. I buy a lot less new stuff than I used to.
posted by fedward at 3:34 PM on June 3, 2015


If I'm going to buy the track I want lossless because there's no patent, transcoding or archival issues (MP3 in the US could still be subject to submarine patents for like two more minutes or something), but I'm not too hung up on quality issues. Even if there is some quality difference, it's not as bad as the problems introduced by my relatively cheap audio setup. That said, I never criticize someone else for wanting "higher quality" even if I can't perceive it. I had an argument with a guy who set up streaming media for the website of a company I worked for. I found the audio so painfully tinny as to be unlistenable, but his concert wrecked ears couldn't hear the difference.
posted by BrotherCaine at 3:52 PM on June 3, 2015


1/6 on NPR using Sennheiser 439s straight from the computer. I tried ABX testing years ago back when I had nothing better to do than dream about being an audiophile nerd and basically determined that couldn't pick anything out. I still insist on FLACs for all my digital music, though (that I then compress to put on my portable).

I remember reading somewhere that a mere 5% loudness difference was far more discernible than basically any encoding for the normal person on normal gear. That always made a lot of sense to me.
posted by ropeladder at 4:03 PM on June 3, 2015


I've gone into the mixing stage for films where in the sound edit we've temporarily cut in msuic files using only 128-320 kbps mp3s. We then receive a WAV file during the mix and are able to A/B them. Holy shit, the difference can be intensely pronounced in that scenario. Not always, and sometimes the 320 is very close, but you can almost always hear it.

Then again, who has a mixing stage to hang out in and listen to music on? In casual listening, for me, 320 is usually fine. The thing is, and I highly stress this, (and forgive me if someone else has brought it up earlier) - very often the MP3 master of new or remastered music is different than the vinyl or CD master. Brickwalled, loudness war, etc. The waves that are used for the vinyl master will have a lot more dynamics. CD's technically have more dynamic range. MP3's aren't too shabby either. But for a number of silly reasons, the files used for mp3's won't get the same dynamic range as the vinyl master. (Sometimes both CD and MP3 get the brickwall treatment, and only vinyl gets the wider dynamic range). The difference between them is pretty clear - it's not a symptom of the format itself, but of the intended use and expectation of audience taste that correlates to the format. So for example if you rip a wave file of a popular song from the CD, convert it to 320 mp3 yourself and compare it to the mp3 downloaded from iTunes, it's possible that it might sound very different, while your own original wave and 320 mp3 might sound very similar.

So I end up buying records that come with free MP3 downloads and yet still go to a torrent site and pirate 24bit FLAC Waves ripped from the vinyl on someone's nice HiFi setup... Because the downloaded mp3's often sound flat.

On this test I listened with nice Beyerdynamics DT 770's straight from the questionable headphone jack on my macbook pro: 0 out of 6.
posted by SmileyChewtrain at 4:48 PM on June 3, 2015 [1 favorite]


Hmm, 3/6, and 2 of them were the 320 mp3s. But one of them was the 128k, so :(

But on the other hand, it was the Katy Perry, and, um, I thought it was supposed to sound like that, I guess? But I had that problem with everything except the classical- everything sounded different, but only a few of them sounded 'wrong'.

I took the test using a macbook pro > (basically a next-gen T-amp that's a factory prototype of something that'll never get made) > some rebuilt Klipsch KG4s. Which is where I do almost all my listening so no excuses there.

My hearing should suck, from years of sticking my head next to speaker cabinets at rock shows, and I guess it kind of does. But I'll still download FLACS by choice, because hey why not. But I'm too old-skool to trust streaming services anyway.
posted by hap_hazard at 5:10 PM on June 3, 2015


If there's one thing that working professionally as a sound designer for live performance has taught me, it's that not even 1% of punters are audiophiles - and that's from a base of people who love and care about the arts. The audio fidelity they listen for - if they listen for it at all - is not obviously shit.

I was raised in an audiophile household, and my Dad's love of great music (which we don't necessarily agree on, actually) is a big part of how I got to be where I am. But Dad's in his 60s now, and his hearing is going - he has hearing aids now - so what's important for him is hearing anything at all. And a few rounds working with deaf dancers, or inclusive companies with disabled and not-disabled performers, or even hearing aid induction loops for a regular show can give you a pretty... pragmatic perspective on what needs to be done.

The other thing I think about - I happened to be talking to a colleague the other day about this - is that there are whole sound cultures for which fidelity isn't even a concern. I'm thinking of the extraordinary Music From Saharan Cellphones kind of world, or when grime and dubstep were being propagated by ringtone in the early years of this century. I first heard 'Pulse X' as a 96kbps mp3 hoisted off Limewire or some such, and it was a revelation.

I get that audiophiles like their Mark Knopfler concept albums to be perfect in every way, but holy lord jeebles perfection is boring...
posted by prismatic7 at 5:24 PM on June 3, 2015 [2 favorites]


I only listen to music the way it's meant to be heard: via uTorrent's built-in media player.
posted by turbid dahlia at 5:40 PM on June 3, 2015


5/6 - I think the Neil Young track's sort of "thick" 70's production fooled me, and I picked the 320 mp3 there. M Audio BX5's (admittedly "studio" monitors, albeit fairly budget ones) just fed directly from my iMac's headphone output.

I guess I get to keep my job. And my username.
posted by soundguy99 at 5:57 PM on June 3, 2015 [1 favorite]


the files used for mp3's won't get the same dynamic range as the vinyl master. (Sometimes both CD and MP3 get the brickwall treatment, and only vinyl gets the wider dynamic range

Do you know to what extent masters for lossy distribution are actually different from CD masters (lossless digital distribution would presumably be just fine using the CD one) now that there is such a thing? You've got me wondering. I see a lot of stuff thrown around about how you should make sure your raw file is not peaking above -0.2/-0.5/whatever dBFS or leave more dynamic range or whatever before feeding it to the MP3 encoder because the conversion process can actually clip it - I think more accurately it would be creating something like intersample peaks - or reduce the dynamics further. Or a lot of MP3 encoders at lower settings do a hard cut of content over 16KHz. So there's also a whole debate over whether its a good idea to preemptively lowpass at 16K when uploading to something like Soundcloud.

By the way an interesting thing in discussions of MP3 quality is how... not completely standardized the standard is. For example while the standard comes with a suggestion for the psychoacoustic model - the algorithm that does the bit everyone always explains where it figures out what information is most important to retain at a given time from the standpoint of human perception - the good encoders like the Fraunhofer one and LAME use different ones.
posted by atoxyl at 6:54 PM on June 3, 2015 [1 favorite]


Most of the advanced lossy compression methods are like that: the standard really only mandates how to decompress the data. It's because as the compression formats become more powerful they also become more flexible, which means that the encoder has to make many more choices. Imagine a hypothetical "photoshop compression" which would represent an image as a series of photoshop operations starting from a blank canvas (e.g., draw a line of this color from point 1 to point 2, copy this rectangle, blur this selection, etc.). Then the decoder would be straightforward to describe and implement (it would be photoshop), but there wouldn't be one true way to implement the encoder.
posted by Pyry at 7:09 PM on June 3, 2015


the files used for mp3's won't get the same dynamic range as the vinyl master. (Sometimes both CD and MP3 get the brickwall treatment, and only vinyl gets the wider dynamic range

Do you know to what extent masters for lossy distribution are actually different from CD masters (lossless digital distribution would presumably be just fine using the CD one) now that there is such a thing?


I can only speak to my own experience at having an album mastered and being mildly surprised (but not shocked) to see two sets of waves come back and the learning it's fairly common. I do know there's been a trend of "mastered for iTunes" in the past couple years - they made a big thing of it a while back but then I think it just faded away into being "standard," to whatever extent. I don't actually know the numbers on it - just that it does happen.
posted by SmileyChewtrain at 12:28 AM on June 4, 2015


"I'm thinking of the extraordinary Music From Saharan Cellphones kind of world, or when grime and dubstep were being propagated by ringtone in the early years of this century. I first heard 'Pulse X yt ' as a 96kbps mp3 hoisted off Limewire or some such, and it was a revelation."

Thanks for those links! I got to listen to both while in a moving bus. And now I'm going to occupy the rest of my commute by digging around for similar stuff on some streaming service.

There are people who think that the equipment I'm using for this is "crap". I feel like I live in the future, and it's awesome. I'd rather have this than a flying car anyway.
posted by bfields at 6:23 AM on June 4, 2015 [1 favorite]


I made a mental note to revisit this on two different set ups, since when I listened at work from the audio out on my Mac into a pair of Audio-Technica headphones, I got 1/6 correct and my results were basically random.

Anyway, I just listened on a DAC with my trusty Sony monitors and got 5/6 (I chose the 320 on the Suzanne Vega -- anyone with a better ear than me know what the tells are there? To my ear it seems like that song has exactly the sort of frequency characteristics that MP3 excels at representing with minimal apparent loss, so I was pretty stumped). I'm wondering which component to blame more for my first result, the onboard DAC or the headphones (which, in their bass heaviness, do not have anything like a flat frequency response).
posted by invitapriore at 6:24 PM on June 14, 2015


Listening to the Suzanne Vega song samples again, for me it was the "sense of space" rather than any specific frequency characteristics. Which is to say, I can hear some reverb (which is almost undoubtedly the microphone picking up the sound of the room she's in, rather than anything added afterwards), and in the uncompressed WAV section you can hear just a touch of this all the time, right from the start. Whereas in the mp3's you can only hear the reverb when she hits a word juuuuuuust a little harder ("diner", "counter") and there's a little space afterwards for the reverb to bloom. So even before i went back and listened critically, I think I picked up on the WAV sounding more "natural."
posted by soundguy99 at 7:59 PM on June 24, 2015 [1 favorite]


« Older Wrestling an Electric Eeel   |   "Today is the end of sheloshim for my beloved... Newer »


This thread has been archived and is closed to new comments