Higher Sample rates better than 44.1/16? Guess again...

[loudboy]

In this month's Mix, the Insider Audio column is about a ABX study that two well-respected audio guys did.

The findings:

None, of the dozens of people tested, in hundreds of trials, on several different systems, could tell the difference between standard CD and SACD or DVD-A formats.

They used a wide variety of people, from music students to pro engineers. They used a variety of systems, but all were very good and tightly controlled. They tested their hearing at 15K and up, but those with the best HF hearing did significantly worse in the test.

The other interesting finding was that all the sources that were originally on higher res sources were thought to sound better, at either sampling rate. This was attributed to the fact that more care was put into the production, seeing as it was going to a higher res final project.

Loudboy

[justicetones]

I could easily agree with the results.

I would even go farther to say that in recording the sample rate has little if nothing to do with the final products end results. I think that a well engineered project can benefit some from a significantly higher sample rate (Not 44.1 to 48) i.e. 24/44.1 to 24/88.2, 24/96, or higher. But the engineering is far more important than sample rate.

Furthermore, I think more harm is done in the down conversion from 48 to 44.1 to be of any benefit.

I still believe that the care taken in recording or the vibe in the studio among personalities has far more effect on the result.

Record on and be creative. :BEER


Great post Loudboy. :BEER

[LSchefman]

>>Furthermore, I think more harm is done in the down conversion from 48 to 44.1 to be of any benefit.<<

Unless you're scoring to picture, where 48 is the standard.

Very interesting post!

I'll have to read the article. I must say I'm not surprised by the results.

[SBRocket]

Well I disagree, but that's what makes this fun I guess. I think 24bit recordings sound different (IMO better, more like the source) than 16bit recordings of the same signal. I also think that 96k sounds different in the same way than 48k does, although I do think that sample rate is not as important as bit depth. That is to say that 96k or 48k makes less of a difference to me than 24bit or 16bit.

I think the problem with these tests as most often performed is that the subject rarely hears the source, just the recordings. I have not read the Mix article so I don't know if that was the case in these tests. But when I hear the source, a 24/96, and a 16/48 I hear the differences. And sometimes they are not so subtle.

It's interesting to me that they claim no one could tell the difference but also say that people thought the ones recorded at the higher sample rates sounded better. What does that mean? Were people able to pick out the "better" ones or not? It sounds like they were. And attributing that to more care being put into the recordings sounds like hogwash to me. I personally have never tried to do less than my best because the recording equipment wasn't the best out there. In fact the opposite might be more true, when I know I have a crappy old deck to record on that might stop working at any time or need an mid-session alignment, I am usually more prone to take it slowly and put in a bit more effort if just to get the sound down safely. Although I do not think that impacts the quality of the work, it just makes it slower.

Although, I firmly believe that it is much more about the recordings and who is making them than sample rate or bit depth. If someone asked me, "Should I get a better engineer or go 24/96?" my answer would always be the former. I also think that the best engineer and the best equipment in the world means nothing if you don't get great performances. There is an element of magic that makes all of this stuff moot anyway. If it's there it's great. If it's not, no amount of knob spinning will put it there.

Finally, as far as downconverting goes, try Barbabatch. It really does a great job.

Steve

[justicetones]

Originally Posted by LSchefman

>>Furthermore, I think more harm is done in the down conversion from 48 to 44.1 to be of any benefit.<<

Unless you're scoring to picture, where 48 is the standard.

Very interesting post!

I'll have to read the article. I must say I'm not surprised by the results.

Les Of course. You are 100% right. Then one must remain compatible with the standard. I was assuming that we were only talking about music for listening. You know what they say.

When picture or commercials are involved 48 is the standard.

Chris

[elambo]

When we first got our dithering box 10 or so years ago, we'd test our ears by switching between 16 and 24 bit to see how well it worked. One person would toggle the BYPASS button and the rest of us would try to guess which we were hearing - 16 or 24. It was actually pretty easy to hear the difference and we rarely guessed incorrectly. That was 10 years ago when dither wasn't as good. Now, especially with pow-r dither, 16 bit can sound nearly indistinguishable from 24 (or as the article suggests - entirely indistinguishable).

44.1 vs 96??? That's an old chapters in the book. I primarily do film and tv which is either 48 or 96 so I'm sticking with them, but I personally hear an improvement in the frequency extension of cymbals (for instance) at higher rates. More air. More open. I haven't read the Mix article yet, but it seems to throw that theory out the window. Placebo then? I really don't think so in this case. I wonder what kind of material they were listening to.

[LSchefman]

I went and read the Mix article, which is not the study, it's a summary of the study. But it was still interesting as hell, I appreciated the author's insight and objectivity, and it was well-written.

So what does that mean for me, in my studio, for my own work? Not much at all. I'm sticking to the way I do it now, because it works fine.

I'll continue to record at 24 bits, because why not? It doesn't tax the computer, and it certainly doesn't hurt the signal, and 48K because that's the standard for video, which is what I write and record music for, and frankly, it sounds just peachy to me.

I could say that I should record at the very highest sample and bit rates for posterity, but I seriously doubt that the television commercials and other media that I score for will be listened to in the future for any reason at all, ever, unless archaeologists can't dig up anything else from the 20th and 21st centuries and consider my work a major "find". And if that's all they have to go on, I feel really, really sorry for them ("gosh, these ancient people never listened to anything longer than 30 seconds!").

No one on earth will be listening to my current work, not even as a joke, 14 weeks from now, let alone 100 years from now.

This is why I don't worry about gear any more. I'll record with anything. I recorded my entire last commercial with a string and a paper cup, and it sounded just fine on an mp3...

[loudboy]

Originally Posted by SBRocket

I think the problem with these tests as most often performed is that the subject rarely hears the source, just the recordings.

They were only testing people's ability to hear the difference between 16/44.1 and SACD or DVD-A. No one could.

The test was ABX, matched to 0.1dB. You know, a real test. <g>

Originally Posted by SBRocket

It's interesting to me that they claim no one could tell the difference but also say that people thought the ones recorded at the higher sample rates sounded better. What does that mean? Were people able to pick out the "better" ones or not?

It means exactly what they said. Overall, the high-res recordings were more pleasing to them, but they STILL could not tell the difference between 16/44.1 and high-res versions of them. The only reason they could come up with was that they were engineered better.

Read the article...

Loudboy

[loudboy]

Originally Posted by LSchefman

And if that's all they have to go on, I feel really, really sorry for them ("gosh, these ancient people never listened to anything longer than 30 seconds!").

"Hot Pockets!"

Loudboy

[loudboy]

Originally Posted by elambo

I wonder what kind of material they were listening to.

"classical instrumental, choral, jazz, rock and pop, from audiophile labels like Mobile Fidelity, Telarc and Chesky."

Loudboy

[justicetones]

Man I need to read this article. I always record at 24 never at 16. But I am very intrigued by the Mix experiment.

[SBRocket]

Originally Posted by loudboy

They were only testing people's ability to hear the difference between 16/44.1 and SACD or DVD-A. No one could.

The test was ABX, matched to 0.1dB. You know, a real test. <g>

Still my point is that without a reference to the source in one's ears, it would seem to me to be more difficult to discern the differences. Sounds more like what? Or less like what?

Originally Posted by loudboy

It means exactly what they said. Overall, the high-res recordings were more pleasing to them, but they STILL could not tell the difference between 16/44.1 and high-res versions of them. The only reason they could come up with was that they were engineered better.

But if they liked them better, does that not prove that they heard a difference? They CHOSE the higher rate recordings as better. They did hear a difference between material originally recorded at higher rates and material recorded at lower rates. It seems they answered the question as true or false not multiple choice but they made the right choice. Maybe it was in the way it was asked.

Originally Posted by loudboy

Read the article...

I will. And I'm sure it's interesting and reflects the research as it was conducted, but my personal experience tells me something other than the way you described the conclusions of the article. But I've been on this side of this discussion before. I'm sure it will be a good, informative read.

No problem.

Steve

[Bassomatic]

Originally Posted by SBRocket

Still my point is that without a reference to the source in one's ears, it would seem to me to be more difficult to discern the differences. Sounds more like what? Or less like what?

I don't see your point at all. The differences are *subtle*. If anything, the recorded samples would sound slightly different from each other, whereas *both* would be relative worlds apart from the live source.

>>but my personal experience tells me something other than the way you described the conclusions of the article.<<

Chances are very good that the study Mix cites was set up a bit more scientifically than impressions gleaned from personal experience, unless that experience includes setting up carefully executed double blind tests and the like.