Question for MEs Re Sample Rate & Sound Quality

[Trakworx]

I was reading the poll/thread "What sample rate do you record at?" in the High End forum, and I thought I'd ask you guys over here in the mastering forum for your opinions. This quote made me want to ask here :

Originally Posted by bluesmanjl

The better it sounds going to mastering, the better it sounds in the end. Ask any great mastering engineer and if they disagree with that, they are not a great mastering engineer period. Every little bit of improvement possible is a big deal to these guys. If they find out there is a first generation digital copy and the original copy might be available they'll be calling and demanding that copy, that's why they're so good , they care. ……… Stay in the best sounding format ………… until the final millisecond.

I have always subscribed to the idea that keeping the bit and sample rate resolution as high as possible for as long as possible would yield the best results in the end, even if it will end up at 16/44 or some other lower resolution. However, I have no data to back this up, it just seems logical to me.

So far, 71% of Gearslutz in the poll say they record at either 44.1 or 48k. So...

Leaving out Hard drive space and DSP usage considerations, just considering sound quality, are they losing something?

If we are recording for CD/MP3 release, should we bother with recording at higher sample rates?

Sorry if this has been asked here before. I didn't find it in my search. Thanks!

.

[mcsnare]

I think the quality of the converter is more important than sample rate. In other words, I 'd rather hear a 44.1 capture from a Prism than a 96K capture from a Digidesign. If all things are equal, MAYBE a higher sample rate will sound better. It's one of those things that is hard to setup a good test, at least in a multitrack situation.

Dave

[aleatoric]

In my opinion the advantages of higher sample rates (88.2k, 96k) come as a result of better DSP performance (aliasing spread outside of audible ranges as an example) and better ADC performance with some converters. If one were to record and mix at 96k there is a chance that it will sound better than if they did at 44.1 do to the reasons above.

As far as sheer sound quality, that is a tricky one (and gets debated a lot). I'd recommend taking a file you received natively at 96k and SRC it to 44.1k (using a high quality algorithm) and do an ABX test and see if you can consistently pick out the 96k version from the 44.1k version.

[jayfrigo]

If 64k was an option, that would probably be my choice, but barring that, 48k will suffice in most cases for initial A to D. However, it's a more complex problem than that. Non-linear processes in DSP are often aided by higher sample rates, so if you are planning to mix in the box with lots of dynamics processing, and you know those plug-ins don't upsample internally (most don't), you may consider a 2X rate. If I'm doing a minimalist project in classical, folk, jazz etc., I may consider recording at a 2X rate as well, but for most large rock and pop productions, 48k gets the nod.

44.1 has enough trouble that the extra space afforded with 48k actually can help, and most times mastering will include a pass through analog, so the target rate is not really that much of a concern. A typical mastering chain will do D to A at whatever rate the source is, go through the analog chain, and then can capture at whichever rate is preferred. In my case, that's 88.2 at the end of the chain so that the limiter is 2X, and then a trip through a hardware Lavry synchronous real-time SRC and into Sonic at 44.1, the target rate for CD. If I'm doing a DVD project, then A to D is 96k so the target is 48k. If doing both, a separate pass at each rate is preferred, but if not possible due to time and budget constraints, modern SRC has no trouble making a very clean conversion, if you choose your SRC wisely of course.

[adamgonsa]

As one of the little steps that can cumulatively add up to a big difference, a higher sample rate can indeed help. As mentioned above though, it might be a waste of space if the source ADC or clock is crummy.

If we are recording for CD/MP3 release, should we bother with recording at higher sample rates?

I think the "a likely output format will be of low quality so why should I bother" line of thought is fraught with problems. Just pretend that everything will be played back lossless on a good system.

44.1 has enough trouble that the extra space afforded with 48k actually can help, and most times mastering will include a pass through analog, so the target rate is not really that much of a concern. A typical mastering chain will do D to A at whatever rate the source is, go through the analog chain, and then can capture at whichever rate is preferred. In my case, that's 88.2 at the end of the chain so that the limiter is 2X...

My preferred capture rate at the end of the chain (for a CD) is 88.2 also.

[macc]

Sort-of-OT, but I generally wince when I see a track has come in at 96kHz... Usually (here, anyway) they are the worst mixes.

[aleatoric]

Originally Posted by macc

Sort-of-OT, but I generally wince when I see a track has come in at 96kHz... Usually (here, anyway) they are the worst mixes.

I wince when I see 16/96!

[Trakworx]

Thanks for the answers!

Mostly I'm seeing answers here like:

"MAYBE a higher sample rate will sound better",

"If one were to record and mix at 96k there is a chance that it will sound better than if they did at 44.1",

"48k will suffice in most cases for initial A to D".

I'm not seeing any strong agreement with the quote in my OP. And he seemed SO confident about it!

Anyone else?

.

[Trakworx]

Originally Posted by jayfrigo

44.1 has enough trouble ...

44.1 was by far the most frequently used SR in the poll with almost 47% of Gearslutz recording that way, almost twice as many as use 48kHz. So...

What trouble?

.

[Adam Dempsey]

Originally Posted by aleatoric

In my opinion the advantages of higher sample rates (88.2k, 96k) come as a result of better DSP performance (aliasing spread outside of audible ranges as an example) and better ADC performance with some converters.

I'd have to agree, and from the start it comes down to converter quality.

Of course one thing to bear mind is that 44.1k really is relevant only for compact disc and medium-low bitrate lossy encoding. If a project comes to me at 48k for online or video release then there really is no need to SRC to 44.1k.

And since shifting studios 4 years ago I think I've only ever had a single project at 192k, fwiw.

[Table Of Tone]

Originally Posted by macc

Sort-of-OT, but I generally wince when I see a track has come in at 96kHz... Usually (here, anyway) they are the worst mixes.

That's often true because a lot of stuff has been recorded with conversion that's not working to it's full dynamic range when used at 96k.

There's also clocking and accuracy issues.

[mcsnare]

Originally Posted by Table Of Tone

That's often true because a lot of stuff has been recorded with conversion that's not working to it's full dynamic range when used at 96k.

There's also clocking and accuracy issues.

I think not. It's more often true because beginners that don't have much mixing experience read on the internets that 96K is better.

Dave

[Table Of Tone]

Originally Posted by mcsnare

I think not. It's more often true because beginners that don't have much mixing experience read on the internets that 96K is better.

Dave

That as well for sure!
Having said that, I've had stuff in from some great ITB mix engineers at 96k that didn't sound as good as other stuff they'd sent me at 44.1.
Mainly digi 192 stuff.

[jayfrigo]

Originally Posted by Trakworx

44.1 was by far the most frequently used SR in the poll with almost 47% of Gearslutz recording that way, almost twice as many as use 48kHz. So...

What trouble?

The filter issues mainly. The steepness and proximity to the desired passband makes it near-impossible to accomplish without artifacts reaching down into the audible range. Eventually you need to get it to 44.1, but for 1X rates, I'd rather record and mix at 48 and keep it away from 44.1 as long as possible. Especially for a DAW feeding a console or summing box for an analog or hybrid mix, there's no reason to do the initial recording at 44.1.

The potential aliasing problems with non-linear DSP will be worse at 44.1 too, so if you're doing an in-the-box mix, that's even more reason to stay higher than 44.1. And, as I said before, since higher-end mastering is usually going to have an analog chain, the target rate is divorced from the source rate since you can just capture at the end of the chain at whatever rate you choose.

However, I can see how a home studio project may want to stay at 44.1 the whole time since there are no analog mixes, analog mastering etc. You just mix back into the box at the same rate, pop a limiter plug-in on there, and burn a CD. That's a different circumstance. It's mainly convenience, but also may avoid bad stock SRC if you don't have a third party SRC that you know is clean.

[Table Of Tone]

The 96/24 mixes I do get in that sound good, tend to be SSL, recorded onto a second PT system, through a high end ADC.

There are a few great mix engineers that work that way and most of em capture at 96/24, regardless of the original SR of the recording.

I admire those guys as they can make an OK recording sound really good!

Obviously, they have great skill, great ears, and their choice of tools is also spot on!

[jackthebear]

Originally Posted by mcsnare

I think the quality of the converter is more important than sample rate. In other words, I 'd rather hear a 44.1 capture from a Prism than a 96K capture from a Digidesign. If all things are equal, MAYBE a higher sample rate will sound better. It's one of those things that is hard to setup a good test, at least in a multitrack situation.

Dave

Totally agree with you there Dave......people easily get bamboozled with the numbers......give me a quality 16 / 44.1 file over a $hitty 24 / 96K one anyday....

and VOTE FOR BANGO!!!

[brethes]

Originally Posted by macc

Sort-of-OT, but I generally wince when I see a track has come in at 96kHz... Usually (here, anyway) they are the worst mixes.

Often true, here as well...

[PBM]

Originally Posted by jackthebear

and VOTE FOR BANGO!!!

... amusing little fellow wasn't he?

Cheers,

Eric

[jayfrigo]

Originally Posted by Table Of Tone

The 96/24 mixes I do get in that sound good, tend to be SSL, recorded onto a second PT system, through a high end ADC.

There are a few great mix engineers that work that way and most of em capture at 96/24, regardless of the original SR of the recording.

In the case of a mix off the console, 96/24 or 88.2/24 is a great way to go, but I bet most of the source sessions (if rock, pop R&B etc.) were probably 48k, or some 44.1.

At some point, computer power and storage will probably make it a moot point, but for now 48k tracking still makes sense, and I don't think most could tell the difference in a majority of cases (48k vs. 96k tracking). Folk, classical, acoustic jazz and other minimalist stuff may be the exception, but even then I wouldn't lose sleep over a well done 48k tracking session.

[Table Of Tone]

Originally Posted by jayfrigo

In the case of a mix off the console, 96/24 or 88.2/24 is a great way to go, but I bet most of the source sessions (if rock, pop R&B etc.) were probably 48k, or some 44.1.

At some point, computer power and storage will probably make it a moot point, but for now 48k tracking still makes sense, and I don't think most could tell the difference in a majority of cases (48k vs. 96k tracking). Folk, classical, acoustic jazz and other minimalist stuff may be the exception, but even then I wouldn't lose sleep over a well done 48k tracking session.

Many of em turned out to be tracked at 48.

I agree that tracking at 48 probably yields the best of both worlds.

[Jeff Hayat]

Anyone see this?

DAW Wars

"16 Bit @ 44.1 kHz (CD standard) is indistinguishable from 24 Bit @ 192 kHz in a sample of over 550 listeners."

Just wondering if anyone had any thoughts on that comment, and the article in general.

Cheers.

[jackthebear]

Originally Posted by PBM

... amusing little fellow wasn't he?

Cheers,

Eric

Indeed.....a breathe of fresh air......

[huejahfink]

Originally Posted by Jeff Hayat

"16 Bit @ 44.1 kHz (CD standard) is indistinguishable from 24 Bit @ 192 kHz in a sample of over 550 listeners."

A statement like this on its own means absolutely jack sh*t.
Where were the tests conducted? How were the tests conducted?
On what equipment were the tests conducted?
Where did the sample come from, and what are their backgrounds?

.... I bet you could take almost any one of those mystical 550 listeners, sit them in front of some 802s in a nice room, and flick between a CD and a high rate file of well recorded jazz or classical music and they would go "oh yeah!"

[mcsnare]

Originally Posted by Table Of Tone

The 96/24 mixes I do get in that sound good, tend to be SSL, recorded onto a second PT system, through a high end ADC.

There are a few great mix engineers that work that way and most of em capture at 96/24, regardless of the original SR of the recording.

I admire those guys as they can make an OK recording sound really good!

Obviously, they have great skill, great ears, and their choice of tools is also spot on!

That is an example of what I was talking about re: it's hard to test. Tone, you hit the nail on the head.
You can't separate the mixer's skill, the SSL, the 'good' converters, ect. from the fact that it's 96K. Have them put down 2 passes next time at 44.1 and 96 and then you might have something to really make an evaluation with.
I'm not saying 44.1 sounds better or worse, just that random empirical evidence is not a good way to come to a conclusion.

Dave

[Adam Dempsey]

Originally Posted by huejahfink

A statement like this on its own means absolutely jack sh*t.
Where were the tests conducted? How were the tests conducted?
On what equipment were the tests conducted?
Where did the sample come from, and what are their backgrounds?

.... I bet you could take almost any one of those mystical 550 listeners, sit them in front of some 802s in a nice room, and flick between a CD and a high rate file of well recorded jazz or classical music and they would go "oh yeah!"

Well, it was only 60 listeners; women "were involved in less than 10 percent of the trials"; and, albeit blind tests, they were structured "to determine their response to signals above 15 kHz" (rather than depth/dimension/detail afforded by higher res...).
Scroll down to 'The Trial' at: The Emperor’s New Sampling Rate -- Are CDs Actually Good Enough?

[Bob Olhsson]

I always tell people to "do your homework and then do what sounds best with your gear and your software."

Implementation makes a much bigger difference than the numbers which at best only indicate there might be a potential for improved quality.

[Jerry Tubb]

Originally Posted by Adam Dempsey

The Emperor’s New Sampling Rate -- Are CDs Actually Good Enough?

(Adam you already know this stuff)

Above 15k is more of a sensation than a "pitch", a bit like a pin prick size air hole.

The other day I played some high freqs for a client 15k-20k, he could "sense it" on the big Dunlavy's.

On ear buds probably not.

Musically speaking 10k-20k is just One Octave, as is 125Hz to 250Hz.

CDs are good enough for most folks, but then there are those of us that want more.

Oliver Twist - Please Sir, I Want Some More | SPIKE

more digital bandwidth, more analog, more everything, except lo-res crap.

I like it all; from Edison cylinders to 1/2" analog tape to 192k digital and beyond.

JT

[krytikal1]

Originally Posted by Bob Olhsson

Implementation makes a much bigger difference than the numbers which at best only indicate there might be a potential for improved quality.

You, Jack, Dave and Mr. Table of Tone have it all right here. I can't tell you the countless times I've been blown away by genius using the most obscure, or "impossible to get a good mix with" gear.

Great taste and great skills/experience will trump all no matter what the tools. Of course it's nice to have all the toys to work with, but you have to know how to use them!

K.