Sample Rate Wars

[creatip123]

Monday boredom again. Feeling like wasting a bit of time.

 

So lately I've observed the number of people with the mindset of 'My devices too good, my ears too valuable, my life too short, to listen to anything under 24bit/192kHz craps. It's 24/192 or nothing at all....'. To top all that, even after they got their hands on that high of sample rate files, they're still playing it through the shitty Directsound. 

 

Browsed around the net, and found this website that got nice sample files (song excerpts) of 24/192 sample rate: http://www.naimlabel.com/musicstore-test-files.aspx. Actually there are a lot of sites providing high sampled files, but their legality is, well, *cough*questionable*cough*

 

I downloaded the one titled: 'Super Hi Definition FLAC'. They cut it to the section of the song where it got good/high dynamic range, obviously to show off 'what high sample rate audio can do' Converted that file to MP3 with various sample rate, to see if it really does sound any different. Nothing fancy, just a LAME MP3 decoder, converted with Foobar's native function.

 

Scientifically, are there any differences (between different encoding and sample rates)? Of course there are. Here are the spectrometer analysis of the files:

 

 

 

 

 

The most distinct difference is the cut-off frequency of each sample rates. 

 

 

Practically, are the differences audible? Now this is where it got subjective. 

 

- Some people can hear them (the differences)

- Some people think they can hear them

- Some people can't hear them (including me)

 

Honestly, I'm having a hard time hearing the differences between the original file and the 130kbps file. I'm lost at 190kbps file. That's why I'm always amazed when I see/read people claiming they can hear the differences between a FLAC and a 320kbps MP3. I'm jealous of their ears, honestly. 

 

So here's the test files, for you to try them for yourself: http://s000.tinyupload.com/index.php?file_id=16679656451208076114 (only 11MB), 

 

One of the right and accountable way to compare them is by doing an ABX blind test on them. This nifty little program does the job well:

 

 

The ABX plugin for Foobar2k. All free: http://www.foobar2000.org/components/view/foo_abx. Just load 2 of the audio files I uploaded to foobar, select them in foobar, right click and select 'utilities => ABX two tracks'. It will even keep track of how much you hit and miss, and calculate the probability of you're guessing and being lucky. Ideally this value should be 0%, but I'd say anything below 20% out of 10 or more tries should be good enough. Try it, it's fun.

 

Well, if you can hear the differences, then good for you, I envy you. If you can't hear the differences (like me), then it's also good, you don't have to waste HD spaces for big, high sample rate audio files

[ShadowCaptain]

I think people take it WAY too far with this stuff,

 

If I had like 30 grand headphones I might want super high end FLAC files, but honestly, I am happy listening to spotify on my momentums, sue me

[Majestic]

I think people take it WAY too far with this stuff,

 

If I had like 30 grand headphones I might want super high end FLAC files, but honestly, I am happy listening to spotify on my momentums, sue me

 

I get listeners fatigue very quickly on Spotify. Something about the audio on that service is off, but I don't know why.

[ShadowCaptain]

I get listeners fatigue very quickly on Spotify. Something about the audio on that service is off, but I don't know why.

 

Really? I pretty much have it on none stop

 

although when I get home I play my files through itunes (also sue me)

 

 

I DO notice the difference, I just .. well dont really care, the convenience out weighs the slight downgrade in audio fidelity 

[SirRoderick]

Smashing good post, I'll give this a try when I get home I think.

[RazeG3SG1]

128kbps vs LEGIT 320kbps have pretty big difference from what I can hear. As for 320kbps vs FLAC, I couldn't hear much difference. But the amount of non-legit 320 is just too damn high! I get FLAC files just to be safe.

[Thorium19]

I can tell the difference between a bad mp3 and a good one, but then I'd expect quite a lot of people to be capable of that, however when listening to FLAC and other lossless high bit rate music, I struggle to tell any difference. I suppose that's because I listen to spotify and itunes too much  .

[MeltingPoint]

160 to 192 vs 320. I can hear minor additional clarity and detail in the upper mids to highs. The difference is subtle. I have to concentrate on listening to pick up the differences.

128 vs 320. I can hear the differences much easier. The quality suffers noticeably. I try to avoid listening to 128.

320 vs FLAC. Cant hear any difference.

When I'm talking about 320, i mean proper 320. Compressed from a lossless source.

[RazeG3SG1]

160 to 192 vs 320. I can hear minor additional clarity and detail in the upper mids to highs. The difference is subtle. I have to concentrate on listening to pick up the differences.

128 vs 320. I can hear the differences much easier. The quality suffers noticeably. I try to avoid listening to 128.

320 vs FLAC. Cant hear any difference.

When I'm talking about 320, i mean proper 320. Compressed from a lossless source.

Exactly! I get FLAC because the amount of trash 320 is just ridiculous...

[creatip123]

128kbps vs LEGIT 320kbps have pretty big difference from what I can hear. As for 320kbps vs FLAC, I couldn't hear much difference. But the amount of non-legit 320 is just too damn high! I get FLAC files just to be safe.

 

Yes, that's the real reason to get FLAC files. My experience is the same. I get FLAC mostly because they're most likely recorded from a good mastered source. In other words, the key is how the recording is mastered. Too bad not many people realize this, and think it's all about the sample rates. They hear a good mastered FLAC, then compare it to a bad mastered 320kbps mp3, and say, 'crap, 320kbps mp3 really sounds like shit. FLAC is the only way to go.....'

[Majestic]

128kbps vs LEGIT 320kbps have pretty big difference from what I can hear. As for 320kbps vs FLAC, I couldn't hear much difference. But the amount of non-legit 320 is just too damn high! I get FLAC files just to be safe.

They can be just as bad. I can grab a 128kbps file and convert it to flac

[xtroria]

I think lachlanlikesathing did some blind abx test of him and his friends of 128k aac vs flac and many of them cant or havr to concentrate super hard to differentiate them.

Would you be able to notice the different in normal listening, probably not

[SSL]

The irony is that a lot of people that worry about this might not even getting bit-accurate music if they are playing on a computer.

[creatip123]

I think lachlanlikesathing did some blind abx test of him and his friends of 128k aac vs flac and many of them cant or havr to concentrate super hard to differentiate them.

Would you be able to notice the different in normal listening, probably not

 

That's one of the quizzes on Philips golden ear challenge. 3 files, 2 are the source, 1 is the compressed MP3 (from below 128kbps to 128kbps), pick which one is the compressed one. Lots of discussions on forums of how people fail miserably on that particular test, on source vs 128kbps.. 

[SSL]

That's one of the quizzes on Philips golden ear challenge. 3 files, 2 are the source, 1 is the compressed MP3 (from below 128kbps to 128kbps), pick which one is the compressed one. Lots of discussions on forums of how people fail miserably on that particular test, on source vs 128kbps.. 

 

Truly amusing if you ask me.

[Izay]

Not hijacking, just a quick question regarding Sample rates and bit depth. I just got a new soundcard and after some research people can't seem to agree if I should put it to 24/48 or 24/96 or alternatively if I should flip it back down to 16/44 depending on what I'm doing (music, movies etc)... If I set it to 24/48.. can't I just forget it? Will I hear a difference if I manually go in and change it every time I want to listen to music instead of playing games? 

[xtroria]

That's one of the quizzes on Philips golden ear challenge. 3 files, 2 are the source, 1 is the compressed MP3 (from below 128kbps to 128kbps), pick which one is the compressed one. Lots of discussions on forums of how people fail miserably on that particular test, on source vs 128kbps.. 

 

Given how people could shout "I NEED 24/192 FOR MUH LCD-2" while LCD-2 is very very forgivable to not so greatly mastered files, it's pretty funny.

 

I think 320kbps MP3s that were mastered from a lossless would sound 95-100% the same for people. Even if you could hear the difference, it's more likely you're concentrating very hard, are you even going to concentrate like that every time you're listening to music? Probably not

[SSL]

Not hijacking, just a quick question regarding Sample rates and bit depth. I just got a new soundcard and after some research people can't seem to agree if I should put it to 24/48 or 24/96 or alternatively if I should flip it back down to 16/44 depending on what I'm doing (music, movies etc)... If I set it to 24/48.. can't I just forget it? Will I hear a difference if I manually go in and change it every time I want to listen to music instead of playing games? 

 

Just set it to the max available, but you won't notice either way unless your card is no good.

[creatip123]

Not hijacking, just a quick question regarding Sample rates and bit depth. I just got a new soundcard and after some research people can't seem to agree if I should put it to 24/48 or 24/96 or alternatively if I should flip it back down to 16/44 depending on what I'm doing (music, movies etc)... If I set it to 24/48.. can't I just forget it? Will I hear a difference if I manually go in and change it every time I want to listen to music instead of playing games? 

 

Set it to whatever value you want, but most people set it to the highest. The directsound will automatically upsample/downsample whatever sample rate your sources are to the value you set. 

 

In theory, upsampling/downsampling adds a little processing time lag. In practical, you won't even notice. Well, maybe if you're dangerously low on system resources. 

 

The thing is, directsound itself isn't very good sounding. Most people who already know this, they use Foobar with WASAPI or ASIO plugin to bypass directsound altogether.

[Izay]

Set it to whatever value you want, but most people set it to the highest. The directsound will automatically upsample/downsample whatever sample rate your sources are to the value you set. 

 

In theory, upsampling/downsampling adds a little processing time lag. In practical, you won't even notice. Well, maybe if you're dangerously low on system resources. 

 

The thing is, directsound itself isn't very good sounding. Most people who already know this, they use Foobar with WASAPI or ASIO plugin to bypass directsound altogether.

So people on Guru and Overclock3d mentioned that setting it to 192 provided distorted sound? Guess that's just a problem with their hardware then? 

The driver has options to set lag, it's currently at 20ms, should I mess with it?.  I flipped between 48/24 and 96/24 and I can hear exactly 0 difference so I guess 96/24 seems like a good spot to settle on. I can go up to 192, but I don't see the point. 

[Oktyabr]

I buy cds (and buy studio masters in flac that wouldn't fit on a CD) and rip them to flac because I want a perfect copy of the cd.  This way my new CDs get played once, while ripping, and then are safely stored away for posterity (and legal reasons), in pristine condition. When I set up for a road trip I burn CDs from flac and in theory they sound exactly the same as the original.  This means I could, if I was really bored, rip THOSE cds back into flac and back and forth, many times, without degrading the sound quality.  MP3 is an inherently lossy format.  Try the same experiment (you can skip the CD burning process, just convert to wav then back to mp3 and back and forth) and after only a few iterations of this process the sound quality degradation would be quite apparent.  What's the point of ripping a cd if the process can't be reversed without loss???

MP3s to save space?  That is so last decade ago.  A one TB hard drive is less than $100 right now and should be able to hold close to 100,000 flac files, including album art, folders, etc. with ease.  The $30 64GB card in my phone currently holds over 3000 flac files with several gb of unused space to spare...

[kuhchuk]

I can definitely hear a difference between FLAC and V9-V3 (I only tested V9, V6, V3, and V0 for the sake of time), with a perfect score on the comparator.  At V0 settings, I feel like I can tell the difference, but with the ABX comparator, I only shoot about 50%, so no proof there.  Haven't even come close to 50% with sample rates above CD quality.

[kuhchuk]

I can definitely hear a difference between FLAC and V9-V3 (I only tested V9, V6, V3, and V0 for the sake of time), with a perfect score on the comparator.  At V0 settings, I feel like I can tell the difference, but with the ABX comparator, I only shoot about 50%, so no proof there.  Haven't even come close to 50% with sample rates above CD quality.

I probably should have mentioned that this was with my own set of files, some months ago, and used 5 songs from vastly different genres in the test.