The Optimal Sample Rate for Quality Audio

[Fletcher]

Whoa... what's next? ...human sacrifice, dogs and cats living together – mass hysteria.

From everything I've heard, I have to completely agree with Dan... for years I've been of the opinion that 96k sounded better [sweeter, more musical] than 192.

I don't know if Dan's reasons for what I hear are right or not [and frankly don't care]... but that's what I heard, and I'm sticking to it.

Peace

[-tc-]

Cheers Mr Lavry ... more easy-reading references to point people to are always welcome

[popmann]

Ever notice how all the people who espouse how pointless 88/96 is here don't chime in?

C'mon, Team Nyquist...tell Mr Larvy how he must be fooling himself into believing he hears anything more at 96k than 44.1...and how he couldn't pass a double blind test...that no human can hear a difference...

[SoSueMe]

Originally Posted by popmann

Ever notice how all the people who espouse how pointless 88/96 is here don't chime in?

Since you insist on throwing out a bone, ever notice how all the people that do not believe in the relevance of science are the ones quick to believe what's written on paper and not the real truth? Ever notice how the 96 kHz fanboys are the ones that never participate in relevant scientific testing?

[justynnuff]

I couldn't agree with popman more. Although I don't think those who say they can tell a difference between 44.1 and 88/96 are flat out lying, they are indeed wrong. I'm sure this has been posted here before, and although its not directly related to the topic, I really like this video debunking all those 'ultrasonic ears' out there. Audio Myths Workshop - YouTube

Although, to play devil's advocate, I don't like how the article explains how sampling at such rates is bad because 1) The data files are larger, 2) More processing power is needed, and 3) because it is less accurate. Although the first two reasons are stunningly obvious, the article only contains cute anecdotes (I can relate to walking up hills!) and refers the reader to his paper on sampling theory for why it is less accurate. And although I only scimmed over the sampling theory paper, I couldn't find a direct relation between higher sampling rates and 'less accuracy' (he talks about OPAMPs and cap charge times, but doesn't bridge the gap). In the end, everything on a signal chain is an intrinsic low-pass filter (the mic, the cable, analog components, studio monitors, the frikkin air, your ears), and the arguments made could easily be made for 44.1/48 instead of 88/96. It's a nice paper for a crash course on DSP theory with a focus in audio, and application; but doesn't do a great job of turning people off from using their equipment to its fullest and within-design potential.

Although my DSP theory is a little rusty, that's my 2 cents

[mikeyman]

ever had your ears checked ??If you can't hear the difference i think you have lost your hearing ...not joking alot of us have lost it.

[SoSueMe]

Originally Posted by mikeyman

ever had your ears checked ??If you can't hear the difference i think you have lost your hearing ...not joking alot of us have lost it.

Ever had your placebo checked? Most haven't and are shocked to learn what they thought they were hearing wasn't what they were actually hearing. A small few are humbled by the results and learn from it. The others usually go into denial arguing science while others conjure up endless new theories or arguments as to how it just can't be.

[mikeyman]

well when you listen to a pro recorded CD you are hearing a product that was recorded at a higher sample rate than 44 ..

[Magucci]

Originally Posted by mikeyman

well when you listen to a pro recorded CD you are hearing a product that was recorded at a higher sample rate than 44 ..

Nope. There're many pros who record in 44.1 only.

[Frost]

Originally Posted by mikeyman

well when you listen to a pro recorded CD you are hearing a product that was recorded at a higher sample rate than 44 ..

From whose studio? There are studios using higher sample rates and plenty that aren't. Some of the finest recordings known came from 44.1. Plenty of "pro" studios are 44 or 48

[Magucci]

Originally Posted by awesomenessman

The faster your system sample rate, the less latency in the system, and that's a big advantage.

Agree.

[mikeyman]

Originally Posted by Magucci

Nope. There're many pros who record in 44.1 only.

perhaps i'm wrong

[Magucci]

Originally Posted by mikeyman

perhaps i'm wrong

It's not your fault.
I wish all people who deal with records would make it at least 88.2 or better DSD.

You can't make the whole world perfect, but you can make your own world better.

[popmann]

I couldn't agree with popman more. Although I don't think those who say they can tell a difference between 44.1 and 88/96 are flat out lying, they are indeed wrong.

You mean you couldn't agree with me LESS.

I see no good excuse for producing music in this day and age at less than 88.2. If you can't hear the difference I'd say you're in the wrong business.

[SoSueMe]

Originally Posted by popmann

You mean you couldn't agree with me LESS.

I see no good excuse for producing music in this day and age at less than 88.2. If you can't hear the difference I'd say you're in the wrong business.

It's always the ones that don't put their money where their ears are. Since you say you can easily hear the "cheap one," why didn't you take the test? Surely ANYONE can hear the difference between a cheap Steinberg and a high end Lavry, right?? LOL

Post #180 here: A/D Comparison - Steinberg MR816 vs Lavry AD11

EDIT: And, I invite you to take the next sample rate test I post, which directly deals with this topic. Will you take that one?

[shteyven]

Originally Posted by Magucci

Nope. There're many pros who record in 44.1 only.

+1
There are so many interviews and quotes all over the interwebs that say exactly this. 44.1, 24 bit!

Sent from my GT-I9100 using Gearslutz App

[Matt Hepworth]

I'm not a believer in higher sampling rates. I can't understand why 64kS/s did not become a standard - the math is right. Marketing must not have thought it was a big enough jump above the then current 32/44/48.

If around 60 is ideal, then 64 just plain makes sense. Only a couple audio sequencers and converters ever supported it, though. You can still make SONAR do it, but I think only RME interfaces can currently support it. None of my current gear supports it, unfortunately, so I stick to 44.1.


@popman, if YOU can hear the difference in recording at higher sampling rates, maybe YOU need better converters. ;-)

[ProfessorNick]

I agree that 44.1kHz/24 is great as an end product.
What if you do sound design and take a sound recorded at 44.1 and drop it down an octave for a sound effect? 22.05kHz! No good.
Or what if you have a session recorded at 44.1kHz and then the artist wants to speed it up or slow it down. Re-sampling issues and interpolation. (Please don't anyone tell me that elastic audio/ ableton warp pr other time stretching sounds transparent.)
I think that out industry has always pushed for high quality and it shouldn't stop. Back when people listened on cassettes were studios saying "we'll just record you to cassette because that is the way people will listen to it." No, they went for quality and high fidelity.
It may not be necessary to release 96kHz/24 bit, but we have quad core phones nowadays! I can't imagine that digital audio will always stay in its early development state. (We have really only gone from 44.1/16 bit CDs when digital first was released mainstream, to 24 bit? What about Moore's law?)

[mikeyman]

Alison Kraus last Cd was recorded at 96k ..Won an award for best sound have a listen to those guitars and i'm not gunna second guess Shipley

[popmann]

@popman, if YOU can hear the difference in recording at higher sampling rates, maybe YOU need better converters. ;-)

That's the thing...yes, sure...but I shouldn't NEED to spend thousands and thousands on converters...when I can just run at 88.2. We could all use nicer stuff. I just got some 44.1 tracks from a UA 2192...and, well--it sounds like 44.1. Never said 44 sounds "bad", but if you can't hear that it's missing info...I don't know what to tell you.

You say you're "not a believer in high sample rates"...does that mean you have recorded at 88/96 from start to finish on the project and saw no noticeable improvement? Or...you just are taking it on faith that it doesn't sound different? Tell me it not because you've read around here that it doesn't matter...

Re: "voting" in the Steinberg thing...I remember telling you what you needed to do to highlight the differences...and you said I could do it...I know I can do it...but, I know what the results of it are going to be-did YOU do it? You are the one with the thread trying to prove a little consumer interface has conversion as good as Larvy's. Not me. I just went to do it for you since you brought it up and I happened to be in the studio with some spare time...and you've taken the clips down?

The whole paradigm of how we shootout gear is flawed. Shootouts are one piece of an equation. Forums clips are an absolute FAIL, IMO, for picking gear. Get your ears and hands on it...and I for one, know someone who just ditched 2 or 3 816x's who hated them...for a new Symphony system, I believe. And the little loopback thing put the 816 near the bottom with the Symphony near the top. Is that a bogus test?

I think you should use the gear you have to make music. I'm happy that you're happy with the 816. Really am. It's the right piece for you right now.

If you look around this forum, you will see similar "proof" that you should be able to record with a AxeFX for all Egtr DI...and for vocals- a Peluso 47 into an ART TubeMP into a Soundblaster card at 24/44...and that will be as good as a handwired tube amps mic'd with a 122V and a U47 into a Neve preamp into a RADAR recorder at 24/96. Six of one...half dozen of another, right?

[AB3]

I wonder what sample rate analogue is? Sounds like a weird question - but it seems that scientifically it cannot be infinite - so there must be number of samples per second in an analogue recording. If a sample rate is like a fan - obviously a really slow one will be choppy. At what sample rate is there so many samples that it is as smooth as analogue?

[Frost]

Originally Posted by AB3

I wonder what sample rate analogue is? Sounds like a weird question - but it seems that scientifically it cannot be infinite - so there must be number of samples per second in an analogue recording. If a sample rate is like a fan - obviously a really slow one will be choppy. At what sample rate is there so many samples that it is as smooth as analogue?

Two times the highest frequency you want to capture.

[AB3]

Let me put it another way - If I am recording to magnetic tape - how many samples per second are being created? Are you saying it is half of the bandwidth of the recording - somehow I do not think that is it.

Originally Posted by Frost

Two times the highest frequency you want to capture.

[SoSueMe]

Originally Posted by popmann

That's the thing...yes, sure...but I shouldn't NEED to spend thousands and thousands on converters...when I can just run at 88.2.
We could all use nicer stuff. I just got some 44.1 tracks from a UA 2192...and, well--it sounds like 44.1. Never said 44 sounds "bad", but if you can't hear that it's missing info...I don't know what to tell you.

I would bet that you would have a totally different opinion if someone handed you the very same tracks and told you they were 88.1. It's all about perception. You can't avoid it; you're human. The same goes for the price of the converter. If someone told you the converter was $8K, you would literally "hear" it differently. There's no escaping it, unless you're an alien from another planet. Sorry to tell you this, but that's just the way it is.

Originally Posted by popmann

You say you're "not a believer in high sample rates"...does that mean you have recorded at 88/96 from start to finish on the project and saw no noticeable improvement? Or...you just are taking it on faith that it doesn't sound different? Tell me it not because you've read around here that it doesn't matter...

To really know if there's a non-perceptual improvement, a careful scientific test needs to be set up and conducted in regards to stacking. Otherwise, there is huge room for perception creep.

Originally Posted by popmann

Re: "voting" in the Steinberg thing...I remember telling you what you needed to do to highlight the differences...and you said I could do it...I know I can do it...but, I know what the results of it are going to be-did YOU do it?

I responded by saying anyone could download the files and do whatever they wanted with them; i.e. HPF/LPF, heavy EQ/Compression; whatever. If what you're saying is that I needed to highlight the sonic differences between the converters beforehand, then that's just injecting more bias into the test.

Originally Posted by popmann

You are the one with the thread trying to prove a little consumer interface has conversion as good as Larvy's. Not me.

Wrong. I clearly state in the first post that I am not trying to prove anything. If you would have read carefully, you would have seen that this is not what the thread is about. It goes much deeper than that. Hint: It's more about the realization of the power of perception.

Originally Posted by popmann

I just went to do it for you since you brought it up and I happened to be in the studio with some spare time...and you've taken the clips down?

I explained in detail why this is. Since the results have been revealed, any further testing is considered invalid. Stay tuned for the next test. I'll PM you if you'd like when I'm about to post it.

Originally Posted by popmann

The whole paradigm of how we shootout gear is flawed. Shootouts are one piece of an equation.

Everything in life is flawed to a certain extent. A well-conducted formal double-blind test is by far the least flawed of them all.

Originally Posted by popmann

Forums clips are an absolute FAIL, IMO, for picking gear.

Most are, yes, because the people that post them are far from experts in the field of psychoacoustics. There are ways of doing this scientifically, producing relevant and valid results, however.

Originally Posted by popmann

Get your ears and hands on it

Allowing other human senses into the equation injects substantial bias into what the person is "hearing" from the device. Do you want to know what it truly sounds like from a straight line from your ears to your brain, or do you want your brain to have a clouded interpretation of what the unit actually sounds like due to the many biases that play a part in the resultant perception of the device?

Originally Posted by popmann

...and I for one, know someone who just ditched 2 or 3 816x's who hated them...for a new Symphony system, I believe.

Yes, looks, feel, price, brand recognition, preconceived notions, etc. can all play a part in such hatred. The word "Symphony" does sound a lot better than "816," so it must be better.

Originally Posted by popmann

And the little loopback thing put the 816 near the bottom with the Symphony near the top. Is that a bogus test?

Point me to this test, please. Is it an aural test, or is it based on specs alone? Specs don't mean anything if you can't hear it. I'd like to know exactly how they did the test.

Originally Posted by popmann

I think you should use the gear you have to make music. I'm happy that you're happy with the 816. Really am. It's the right piece for you right now.

You forgot I own Lavry gear. I own the AD11 and the DA11, along with other higher end pieces. I also own tons of low end gear, including two MR816s.

Originally Posted by popmann

If you look around this forum, you will see similar "proof" that you should be able to record with a AxeFX for all Egtr DI...and for vocals- a Peluso 47 into an ART TubeMP into a Soundblaster card at 24/44...and that will be as good as a handwired tube amps mic'd with a 122V and a U47 into a Neve preamp into a RADAR recorder at 24/96. Six of one...half dozen of another, right?

Proof? Who's trying to prove this?

I'm in the camp that believes a good engineer can get killer sound from a Victrola if he/she knows how to work the gear, yes. Will it be as good as the really high end stuff? That's in the ear of the beholder.

I'd rather have a trusted mechanic work on my cars with Walmart tools than a pre-teen with no experience using SnapOn tools.

[elpedro]

to me it's simple i record at 24-bit 48K,because i can here the difference from 44K,i can't here any difference at higher sample rates,so i'd rather have a bit more DSP power natively than process "dog-sounds".call me a simplton if you like.....

[TRJanuary]

Originally Posted by mikeyman

Alison Kraus last Cd was recorded at 96k ..Won an award for best sound have a listen to those guitars and i'm not gunna second guess Shipley

But the CD playback you are hearing is sampled at 44.1, not 96k

44.1 sounds pretty good, huh?

[jupiter8]

Originally Posted by AB3

I wonder what sample rate analogue is? Sounds like a weird question - but it seems that scientifically it cannot be infinite - so there must be number of samples per second in an analogue recording. If a sample rate is like a fan - obviously a really slow one will be choppy. At what sample rate is there so many samples that it is as smooth as analogue?

A very common mistake. The output of the DA converter gets filtered so it is as smooth as analog. So higher sample rates do not result in smoother waveforms,you just can capture higher frequencies.

[MaartenT]

Well, actually higher sample rates DO result in smoother waveforms. At the nyquist frequency (fs/2) a high order filter is placed. This affects the higher frequencies in phase and volume. When you use a higher sample rate, the ripple gets displaced above audible frequencies. As a result the audible range has less distortion

[John Willett]

Originally Posted by Magucci

Nope. There're many pros who record in 44.1 only.

Agreed - I was at a session in Abbey Road not long ago with the LSO and it was all recorded at 44.1kHz onto Sequoia (not by me).