NPR Quiz – can you hear the difference?

[Phil O]

I'm not much of a Neil Young fan, but I have heard Harvest in it's entirety and I've always thought of it as one of the worst mixed recordings of all time. I agree with Shooshie that it sounds phoned in. I guess part of this test is to put yourself in a space where you can ignore the musical production and just listen to the technology. That's not always easy to do, especially when you're a musician.

Phil

[Shooshie]

I think the test was instructive in two ways. First, it demonstrated what some of us have known for a long time: 320kbps mp3 and other high-quality compression algorithms don't necessarily suck the life out of a song. Secondly, it forced us to listen to the actual mechanics of song snippets out of context. By listening to Neil Young without knowing that it's his greatest masterpiece, we hear flaws that suggest that any accolades it received were not helped by the mix.

There's a gripe that I used to have about recording studios and their resident engineers, long before I acquired my own. Some engineers just did the same thing every time on every song, every client. Try to get them to do something your way, and you're likely to hear a litany about all the awards they've won, or the platinum records they've mixed, and how they didn't get those by way of taking advice from some nobody loser "who thinks he knows more than me." The same could be said for some arrangers who add the decorations and filler sounds in a song, often in conjunction with the studio. The result may be a formula for top-40 hits, but it may be completely out of character for the music that was brought to them to record. It reminded me of those old "Music-minus-one" records which you could use to play along and learn a piece of music, or to learn to improvise jazz. "Top-40-Hit-Minus-One" is the same hit formula applied over and over, with someone else's voice and song dropped into it. And maybe there's nothing wrong with that approach if you're recording Tony Bennett, Frank Sinatra, or Dean Martin. People buy those albums to hear those great pipes. Doesn't really matter what the fiddles are doing, or the reverb, or how it was EQ'd or compressed.

Other studios had creative people who tried to hear what you were doing, then attempted to enhance that in the extra arranging and even the mixing itself. The result was something more organic. That's where you got the Sgt. Peppers' albums, the Pet Sounds, Bohemian Rhapsodies, etc.

That said... I agree, Michael Canavan, that we can't judge a record in 30 seconds. But to listen critically to an anonymous snippet, without context, can certainly teach us something.

Shoosh

[Michael Canavan]

I think the test was instructive in two ways. First, it demonstrated what some of us have known for a long time: 320kbps mp3 and other high-quality compression algorithms don't necessarily suck the life out of a song.

Absolutely, at least for me it was hard to tell the difference between the 320 and WAV files, between them I suspect it was all guessing on my part. The 128 stood out of course.

[monkey man]

Ditch the VBR thing, like, yesterday, Shoosh, and you know I rarely, if ever "instruct" in this way.

I was always sus on it purely for the reason that I didn't feel comfortable with the amplitude resolution's being constantly adjusted, if indeed that was what was occurring. It just felt wrong to me that any resolution adjustment was being undertaken, a clandestine movement of the goal posts, if you will, so I opted for 320k "normal stereo" (not the other option) with CBR encoding as the method for my iTunes lib rebuild. It took me around 4 years to complete, but boy, was it worth it!

[Shooshie]

Ditch the VBR thing, like, yesterday, Shoosh, and you know I rarely, if ever "instruct" in this way.

I was always sus on it purely for the reason that I didn't feel comfortable with the amplitude resolution's being constantly adjusted, if indeed that was what was occurring. It just felt wrong to me that any resolution adjustment was being undertaken, a clandestine movement of the goal posts, if you will, so I opted for 320k "normal stereo" (not the other option) with CBR encoding as the method for my iTunes lib rebuild. It took me around 4 years to complete, but boy, was it worth it!

Well, I was an early adopter of LAME, using it back in the mid-1990s to create demos for the web, back when you had to learn how to use it and encode your configuration before running it. Back then, you could roll a demo with VBR and it sounded nearly as good as 320kbps, often identical, but at a fraction of the size. This was very important before everyone was networking at megabytes per second.

Now, I generally use AAC, MP4, or I make an mp3 at the highest resolution, so VBR doesn't really switch in. But I still like the idea that I can barely tell the difference, yet save about 3 bits for every bit encoded. (at high-quality. Low quality VBRs can save many times that.) These days, however, it's hard to beat AAC for quality, and the size at that level is remarkably un-huge.

Shoosh

[Gravity Jim]

Yes, Harvest is a terrible recording.

And yet, it touched more hearts and became a part of more lives than the entire recorded output of all the members of this forum put together. Because music isn't about hearing things,, it's about feeling things.

I like a good sounding recording as much as the next fellow, but... so much for specs.

[Robert Randolph]

Yes, Harvest is a terrible recording.

And yet, it touched more hearts and became a part of more lives than the entire recorded output of all the members of this forum put together. Because music isn't about hearing things,, it's about feeling things.

I like a good sounding recording as much as the next fellow, but... so much for specs.

And yet... this is a topic about judging how a sample sounds, not how many people like it.

[Michael Canavan]

Yes, Harvest is a terrible recording.

And yet, it touched more hearts and became a part of more lives than the entire recorded output of all the members of this forum put together. Because music isn't about hearing things,, it's about feeling things.

I like a good sounding recording as much as the next fellow, but... so much for specs.

And yet... this is a topic about judging how a sample sounds, not how many people like it.

Uh? actually it's a topic about the degradation of a recording brought on by the mp3 format, the side topic of whether a recording has tape hiss and is therefore bad etc. was brought on by you.
So essentially you brought up the topic of not liking the sample, not whether the mp3 was degrading it, but your own distaste in the recording. That's of course fine, but since it was you who first started talking about things beyond the mp3 degradation I think it would be wise not to pass judgement on people for further deviating from that initial topic. What I find interesting is there's a few solid bricks in the samples given, but we're not talking about them.

[Robert Randolph]

Uh? actually it's a topic about the degradation of a recording brought on by the mp3 format.

Which would mean: we're talking about how the sample sounds.

I understand that this phrase could mean multiple things, and that's precisely the point I'm trying to make by saying it.

The test is artificially increased in difficulty by presenting poorly recorded samples and a wide variety of genres. We're being asked to evaluate one aspect of the quality of audio that is coming out of the speakers, but also implicitly being tempted to evaluate other aspects of the sounds if you're unfamiliar with the recording style.

I got by on the test by just listening for specific artifacts that I've learned to listen for. I think very few people have taken the time to do this, and instead are trying to listen for key aspects of the recording that are missing or present. That method requires some level of familiarity with genre and production style. The way this test is setup mitigates these factors, and I think that it unnecessarily creates incorrect responses for the majority of people being tested. Who cares if you can't tell Jay-z 128kpbs and WAV apart if you never listen to hip-hop or rap?

And for continuity, my earlier response in the thread:

In the context of the quiz I feel like it's dishonest for the article to include a lo-fi song when you're being asked to differentiate between high-quality and low-quality encoding. Somewhat ironically though, that's the easiest song for me to pick out.

[Michael Canavan]

Uh? actually it's a topic about the degradation of a recording brought on by the mp3 format.

Which would mean: we're talking about how the sample sounds.

I understand that this phrase could mean multiple things, and that's precisely the point I'm trying to make by saying it.

The test is artificially increased in difficulty by presenting poorly recorded samples and a wide variety of genres. We're being asked to evaluate one aspect of the quality of audio that is coming out of the speakers, but also implicitly being tempted to evaluate other aspects of the sounds if you're unfamiliar with the recording style.

Sure, but to frame the debate about recording quality just doesn't work IMO. Some genres of music work with recoding quality as a tool to create a vibe. Black Metal comes to mind, I'm sure most people don't get it, but some things are intentionally recorded poorly to create a sense of distance in time etc. The original way guitarists got distortion was by abusing their amps and actually poking holes in their speakers. Now even with that in mind, the artists actually intentionally worked to create a certain level of degradation, to create a certain environment. Even with bit reduction distortion plug ins you simply don't get the kind of hiss that aliasing 128mp3's have. Nobody likes that sound and it doesn't represent what the artist wants, so despite your annoyance at badly recorded music being on the list, it's makes total sense for it to be there.

Take into account that because the aliasing was more obvious in tracks with more distortion in your own words the Young track was the easiest to pick out. Simply because white noise from overdriven tape, hiss, etc. exaggerates the aliasing. It's actually worse on more distorted tracks. So it's not really possible to say that they were intentionally making it harder by including distorted tracks.

Out of curiosity I would have liked for them to include a Thrash Metal or Hardcore Punk song in there. I think the result would be we would all hear the degradation on the 128mp3 track pretty quickly. Hard to say though, because a chainsaw guitar is pretty much full spectrum white noise, so it might cover up the aliasing a bit. <unintentional pun there.

[Shooshie]

Yes, Harvest is a terrible recording.

And yet, it touched more hearts and became a part of more lives than the entire recorded output of all the members of this forum put together.

So did Andy Kaufman's video of him lip-syncing Mighty Mouse: "Hear I come to save the day!"

Because music isn't about hearing things, it's about feeling things.

So, let's not confuse quantity with quality. Neil may have touched more lives than we have, but that doesn't mean Harvest is a better album — touches hearts more deeply — than anything we have done. Just that he was in a position to sell more albums.

And not to take anything away from Neil, but you may not realize who all reads this forum, even if from time to time. Some of the best guys in the business use or have used Digital Performer, and have been by this forum to learn something. Maybe even to pass on a nugget or two. Maybe Neil's Harvest is better than anything I've done, but I'm not worthy to touch the coattails of some of the folk who have clicked on this forum.

Shooshie

[mikehalloran]

When I posted this, I expected that most of us would do pretty well. After all, whether we make a living doing this or not, all of us have an interest in audio. As such, we have all made some attempt to educate our ears and brains, consciously or not. The fact that we use DP makes us part of the creative process.

How many of us have listened to the various MP3 resolutions -- especially in the days before cheap storage media -- to determine how good was good enough? Of course we can hear the artifacts -- we know what to listen for.

That test wasn't designed for us. It was aimed squarely at the armchair "experts" who love to pontificate about their golden ears the way that some go on about their wine cellars.

I don't care how many lives were touched by "Harvest". It bored the crap out of me when released and my opinion hasn't changed.

[James Steele]

That test wasn't designed for us. It was aimed squarely at the armchair "experts" who love to pontificate about their golden ears the way that some go on about their wine cellars.

You wouldn't have a list of those people by any chance? I have a bunch of gold-plated, oxygen-free, $2,000 speaker cables I'm trying to unload.

[Phil O]

...The test is artificially increased in difficulty by presenting poorly recorded samples and a wide variety of genres. We're being asked to evaluate one aspect of the quality of audio that is coming out of the speakers, but also implicitly being tempted to evaluate other aspects of the sounds if you're unfamiliar with the recording style...

I read this a couple of times before deciding that I totally agree with it. I think this accurately describes what I was thinking, but didn't successfully articulate.

Phil

[philbrown]

I only listened for a few seconds of each and missed two, both in genres I'm unfamiliar with. The ones I got right, I immediately knew within 2-4 seconds, and the two I missed I was never sure and was basically guessing. To me it's all about the depth of soundstage and to a lesser extent, stereo separation and aliasing. In general the WAV's sound 'deeper' (front to back) and the 128's sound 'flat' (2D as opposed to 3D).