Bit rates & double blind listening tests

[jimtron]

I did some Googling to see which bit rate to rip CDs at--the sweet spot between aural quality and file size. Here's an interesting article about a test where folks listened to a familiar track at different bit rates as well as uncompressed and tried to tell which was which.

So 192kbps VBR seems like a good compromise to me. Anyone else have thoughts on this?

Then at Absolute Sound I found this unconvincing argument, "Blind Listening Tests are Flawed," with a lively debate in the comments below the article (I found the comments by Terje Skjaerpe to be particularly astute.)

 

[TheDaver]

I assume you’re talking about MP3 specifically.

It’s all about what seems like a good compromise to you. The community at HydrogenAudio.org seems to think 192kbps VBR should be indistinguishable from CD to most people, and my experience – with my entire CD collection – agrees with this.

But you should also ask yourself what you’ll be doing with your MP3s. If you’ll mainly be listening to them on an iPod or in a car, then you might be able to get away with a lower bitrate. If you’ll mainly be listening to them on good equipment in a quiet environment then you might want to go higher.

And yeah, that Robert Harley guy sounds like a typical audiophile jackass, grasping for something to justify his expensive compulsion.…

 

[jimtron]

I assume you’re talking about MP3 specifically.

I'm referring to ripping CDs to play on a computer or portable listening device, but not necessarily in the MP3 format--I've been using AAC.

It’s all about what seems like a good compromise to you. The community at HydrogenAudio.org seems to think 192kbps VBR should be indistinguishable from CD to most people, and my experience – with my entire CD collection – agrees with this.

But you should also ask yourself what you’ll be doing with your MP3s. If you’ll mainly be listening to them on an iPod or in a car, then you might be able to get away with a lower bitrate. If you’ll mainly be listening to them on good equipment in a quiet environment then you might want to go higher.

Yeah, 192/VBR seems like the sweet spot for me.

 

[Robo Sapien]

The difference in bitrate is more distinguishable in the lower ranges, so it will become notable if listening on a system with a subwoofer. As TheDaver pointed out, quality gear will make the difference.

 

[kuroyume0161]

Nyquist Frequency. Differences in bitrate will actually be more noticeable at higher frequencies since less and less information can be stored as bitrate/sampling rate is lowered. After you have one point for the crest and trough, pseudo-frequencies become evident (ringing artifacts). Once the sampling is beyond a certain number of times the highest frequency, there won't be much noticeable difference between analog and digital to the human ear.

 

[CORed]

I did some Googling to see which bit rate to rip CDs at--the sweet spot between aural quality and file size. Here's an interesting article about a test where folks listened to a familiar track at different bit rates as well as uncompressed and tried to tell which was which.

So 192kbps VBR seems like a good compromise to me. Anyone else have thoughts on this?

Then at Absolute Sound I found this unconvincing argument, "Blind Listening Tests are Flawed," with a lively debate in the comments below the article (I found the comments by Terje Skjaerpe to be particularly astute.)

Well, his argument against blind tests seems to be, "The results weren't what I expected, so clearly the method, not my preconceptions are flawed."

 

[jimtron]

Well, his argument against blind tests seems to be, "The results weren't what I expected, so clearly the method, not my preconceptions are flawed."

Exactly. He really doesn't make a good case for himself, and of course it's in his interest as an audiophile reviewer to promote expensive gear.

 

[JWideman]

Well, his argument against blind tests seems to be, "The results weren't what I expected, so clearly the method, not my preconceptions are flawed."

More than that. A direct quote:

The answer is that blind listening tests fundamentally distort the listening process and are worthless in determining the audibility of a certain phenomenon.

Put another way, "It only sounds better if you know you're listening to it."

 

[jimtron]

I wonder if knowing who made the component and how much it cost would "distort the listening process" at all.

 

[TheSkepticCanuck]

When I listen to ripped CDs on my computer, with good studio quality headphones, I tend to prefer 256 or 320 kbps CBR MP3s. However, on my phone / MP3 player, I tend to use 128 kbps CBR, as my headphones are not good enough to distinguish anything better than that, regardless of the quality of my ears. As was mentioned above, it all depends on the quality of the equipment it is being played on, up to a point.

 

[Delvo]

I'm not sure the concept of "compromise" really makes sense here. Vast amounts of data storage are so easy to get now, that there's no incentive to tend toward more compression.

 

[jimtron]

When I listen to ripped CDs on my computer, with good studio quality headphones, I tend to prefer 256 or 320 kbps CBR MP3s.

Have you tried a DBT comparing that to say, 192 VBR?

I'm not sure the concept of "compromise" really makes sense here. Vast amounts of data storage are so easy to get now, that there's no incentive to tend toward more compression.

If I'm not mistaken a 192kbps VBR track is about 1/10th the size of an uncompressed WAV or AIFF. I've heard compelling arguments that many--if not most--of us can't distinguish between 192 and higher quality files. I agree that hard disk is cheap, but with thousands or tens of thousands of tunes, why take up 10 times more space than necessary? Also, portable listening devices are capacious, but do have limits.

 

[Rairun]

I can listen to and enjoy 128kbps MP3s. If I don't know the song at all, even 96kbps sounds decent enough. The problem is that as I get to know the song well, I start noticing all sorts of little artifacts that really get on my nerves. If you give me a .wav and a 160kbps sample, I'm sure I wouldn't be able to tell the difference right away. But after listening to the song for a week, I might be able to pass a blind test.

So I encode all my music with LAME V0 VBR, which leaves me with 200-210kbps mp3s. I could probably lower the bit rate without noticing anything, but I'd hate to have to rip my CDs all over again if the assumption turned out to be wrong.

 

[TheSkepticCanuck]

Have you tried a DBT comparing that to say, 192 VBR?

No, I haven't. I actually rip all my CDs to FLAC format, and download audio in FLAC format if available, which gives lossless compression. When I want to store the files on a portable device, I simply use a program to export the FLAC files or higher grade MP3 files to 128 kbps directly to the portable device. That way, I have a high quality source, I can downgrade easily. I know I can't usually detect the difference between the higher bit rates, but I prefer the best quality archives of songs possible, if for no other reason than my own peace of mind, just in case I need a perfect quality source for some reason in the future.

The funny thing is, I am not nearly as anal retentive about video files.

 

[Soapy Sam]

The bit of equipment most critical to audio appreciation is the one audiophiles most commonly seem to forget; the ears.
After 30 years on drilling rigs, mine ain't what they were in my teens- and neither are those of anyone over 40.
The best reproduction in the world doesn't matter a toss if you can't hear it.

 

[TheSkepticCanuck]

The bit of equipment most critical to audio appreciation is the one audiophiles most commonly seem to forget; the ears.
After 30 years on drilling rigs, mine ain't what they were in my teens- and neither are those of anyone over 40.
The best reproduction in the world doesn't matter a toss if you can't hear it.

It also depends on what you are listening to. Compression can do more damage to something like classical music than to rock and roll music. I notice it most on things like piano and orchestrated pieces, where as rock and metal music does not seem to be harmed as much by lower bit rates. This is just my own anecdotal experiences, other peoples' experiences may vary.

 

[Cuddles]

I'm not sure the concept of "compromise" really makes sense here. Vast amounts of data storage are so easy to get now, that there's no incentive to tend toward more compression.

Not true at all. To start with, storage may be easy to get, but it still costs money. More importantly, it's often not at all easy to get vast amounts of data storage in the places people actually want to use it. For example, my mp3 player can only hold 16GB. That's a couple of years old now so I assume bigger solid state devices are available, but not that much bigger. Even with all my music compressed as fairly low bitrate mp3s and oggs, I can't fit anywhere near all of it on there. Use FLAC and I can fit maybe half as much. With no compression at all I could fit only a very small fraction.

There always have been and always will be a lot of incentive to have as much compression as possible. The question is always simply where you are willing to compromise between storage space required and data loss.

 

[Almo]

We did a small blind test. At 128 kbps VBR AAC m4a we couldn't be positive we heard a difference. 68 kbps was noticable, 256 kpbs was indistinguishable from uncompressed wav. This is with a PS3 streaming over HDMI to a $3000 stereo with a sub and tower speakers.

 

[Dr H]

I'm referring to ripping CDs to play on a computer or portable listening device, but not necessarily in the MP3 format--I've been using AAC.

I avoid the whole problem by ripping my CDs to 8-track tapes, to be played in the car.

 

[pipelineaudio]

Compression can do more damage to something like classical music than to rock and roll music.

In a classic double entendre, though I suppose you are speaking of data compression, dynamic range compression acts in much the same way, especially in light of the loudness wars of today...and dynamic range compression can increase the efficiency of data compression and on and on it goes