KVR Audio

Robert Randolph · Post by **Robert Randolph** » Sun Sep 05, 2004 11:11 pm

Actually many converters will sound WORSE at higher sample rates. But apparently people think "oh, it sounds different... it's higher.. that must mean it's better". I think if you did a purely qualitative blind test given the same converter with material that specifically has such high-frequency information Id be interested in hearing a detailed report.

Im of course assuming you're using something like earthworks, tannoy, adam-audio etc.. monitors that have a published range well above 22Khz? And working with samples/microphones that can reproduce such information?

Kingston is of course correct (if not slightly flawed semantically, but eh, Id be no better if I went off on it).

Also robain... most sample libraries will be using up to 100's of notes of polyphony if they were well sampled and programmed. Take for instance a multi-miked piano. Say 4 mikes... overheads, room, pzm. You play a jazz chord and you could easily be eating up 20-30 samples at ONCE. Play another chord while the previous one is sustaining... you're up to 60. Play some intermediate bach and you can get into 100's of notes streaming from disk at once (and ram of course).

That's assuming it's a multi-velocity multi-variation sample library that gives you multiple recording pallete options. Without such basic things I dont think you could even begin to call a sample library quality... and if it's not meant to be high-quality, what's the point of higher sample rates?

I do extend my invitation again though, if there is a great reason for perhaps sync/timing, production methods needing 96khz etc... Please let me know. There's a number of libraries coming from lastufka libraries, and it's quite possible we would take into account that need.

Majken · Post by **Majken** » Sun Sep 05, 2004 11:57 pm

44.1kHz vs. 96kHz is very interesting. I've read up on as much as I could find and both tested and thought about it. My personal conclusion is that the benefits are either extremely small, beyond our hearing range or pure speculations.

The inital statement from companies selling 96kHz gear was how much better it sounded. From bass to treble. With criticism based on nyquist this seems to have changed though. 44.1kHz samplerate can play a perfect 20kHz sinewave which is all we need.

Recently phase differences in stereo separation have been spoken of as 96kHz's real benefit. Companies like MOTU use this as a main selling point.

To me it all seems flawed though, sure 96kHz may give a better treble rolloff if the filter at 44.1kHz is bad. However even where the rolloff starts at semi decent equipment is beyond alot of peoples hearing range. This would also need the chain to be perfect. The combinations of microphones, microphone preamps, recording media, storage media and playback equipment that can even playback 20kHz properly are few. The greatest benefit would likely be in software based music using synths playing square waveforms which give high overtones beyond 20kHz. And phase differences in stereo? If this was true it would mean speaker placement and room design was extremely important. Even at 44.1kHz stereo width is a fragile thing and moving around in a room or sitting at the wrong place will destroy or reduce stereo separation.

96kHz seems more like a reason to sell equipment for me than something we actually benefit from. If there are improvements they're not even close giving the improvements a good eq, reverb, guitar amp and such can give.

I can't help but feel sad for the audio industry on the music side of things. At the moment 24bit 44.1kHz seems to be all we need. Surround didn't work out very well in music mixes either. You look at the movie business and you see the improvements dvd gave over vhs. However that's still pretty low resolution. With larger displays at higher resolution movies could look even better. Surround sound worked out great as well since you sit still while watching a movie, which is not always true for when listening to music.

Ah well just my ideas and opinions. Sorry to steal the thread btw. Don't mean to sound discourageing Robains, I hope you find some high quality, high resolution samples so you can get your own opinion on 96kHz. And thanks Robert Randolph for some interesting reading, we seem to share alot of thoughts and ideas.

/Majken

Jeremy_NSL · Post by **Jeremy_NSL** » Mon Sep 06, 2004 12:05 am

I'd like to see all sample libraries adopt 24-bit. Using complex effect chains, especially with compression, can reveal 16-bit noise floors easily.

But 96khz? Couldn't care less. Basically just a marketing gimmick so far as I can tell - if not, then as Majken says the benefits are "extremely small". And the performance tradeoffs, as RR outlined, are awful.

Of course, many of the pros swear by 24/96 or even 32/192 or some such. But these guys need to justify the $20-30k on their Protools HD systems somehow...

Robert Randolph · Post by **Robert Randolph** » Mon Sep 06, 2004 12:08 am

All the tests Ive found show we cannot differentiate between sine/square/saw/triangle etc... above 15khz. Best case scenario, beyond 18khz (which was an AES paper... good place to start selling new tech).

Of course... that makes perfect sense as well. Why should we begin to think just because it's part of a specific signal that we could magically hear above 20khz.

I do think there's a needed room above our hearing range though. Certainly not as extreme as the industry are trying to tell us. Say 50-60khz sampling rate and monitors that extend to around 25khz. Not specifically because we can hear it...

Of course the real limiting factor in all this is microphones... which are the weakest point in the modern recordists signal chain, no matter how "great" your mic is, it still doesnt even come close to reproducing anything spectacular. Monitors come in a close second... but at least we've come much further in that area

Donnie · Post by **Donnie** » Mon Sep 06, 2004 12:10 am

I'll take this discussion a step further. A lot of what is recorded today in the sampling industry even if it's recorded at 96k doesn't matter anyway. I see gear list's posted all the time with items like "M-Audio", "Rode", "ART", "Presonus" etc. which in all honesty makes recording at anything above 44.1khz a moot point. If your talking about real high end gear "Lavry", "Crane Song", "GML" for example THEN you can talk about higher sampling rates.

I STRONGLY encourage everyone to take the sample libraries you have and open up one or two .wav's in an audio editor and listen to them with headphones on. In 9 cases out of 10 I can guarantee you that you will be shocked at what you hear.

Donnie

Robert Randolph · Post by **Robert Randolph** » Mon Sep 06, 2004 12:15 am

I even wonder how many people can hear above 16-18khz these days anyways with people going to rock concerts, clubs etc.. without ear plugs, monitoring for long times at loud levels, playing loud instruments without protection etc...

I think people are simply fooled by the "bigger numbers are better" syndrome. There's always a point of diminishing returns you know...

robains · Post by **robains** » Mon Sep 06, 2004 2:56 am

Thanks for the responses.

I'm not looking for anyone tell me, or anyone, what I can and can not differentiate -- I'm just looking for quality 24bit 96kHZ samples and I'll decide on what I like and/or don't like.

As far as hardware goes, so far 24bit 96kHz has NOT been a problem for my system at any level of polyphony and/or other complexity.

But please let this thread not go the route of debate about human earing, feeling, etc. etc. because there just is NOT enough quantified/qualified data to make any blanket statements in such an area. Seeing as how music is much a personal experience I don't see how making any of these statements is meaningful.

Now, anyone have some 24bit 96kHz samples so I can decide what is or isn't good for me?

Thanks, Rob.

Donnie · Post by **Donnie** » Mon Sep 06, 2004 3:00 am

I think possibly the better argument is should people who CAN'T hear the difference be pedaling sounds?

Donnie

Innominandum · Post by **Innominandum** » Mon Sep 06, 2004 3:39 am

Have you checked out the Vienna Symphonic stuff?

Robert Randolph · Post by **Robert Randolph** » Mon Sep 06, 2004 4:02 am

Donnie wrote:I think possibly the better argument is should people who CAN'T hear the difference be pedaling sounds?

Donnie

I dont understand this statement...

robains · Post by **robains** » Mon Sep 06, 2004 4:06 pm

I googled "Vienna Symphonic" and unless I found the wrong library, all there samples are 16bit 44.1kHz.

They do sound nice, but not as clean as I was hoping, but this could just be a limitation of listening online.

blixa · Post by **blixa** » Mon Sep 06, 2004 4:54 pm

Kingston wrote:By the way, our ears (ie the hardware) can only detect frequencies and their phases of up to about 20k. Try to restrain yourself from distrubuting any other kind of information!

but my pet bat won't like that!

sorry for this lame joke. seriously, it's a very informative thread.

asseca · Post by **asseca** » Mon Sep 06, 2004 5:15 pm

Have you tried the Sonic Reality samples ?

ChimeraMan · Post by **ChimeraMan** » Mon Sep 06, 2004 5:34 pm

I can't count the times I've seen or heard the 48/96 argument! And as for what the brain/ears/genitals can actually physically acknowledge...does anyone actually f**king care? Correct me if i'm wrong but do we not lose a certain amount of bottom-end the higher we climb the sample rate ladder? Well that's what i've noticed...Surely a lot of this shit depends on how well the individuals' ear is trained to specific frequencies(without stating the obvious)?

All I know is..i'm fed up to the back teeth with that f**king dog barking next door! I can feel a "rat poison through the letterbox" scenario coming on.

cron · Post by **cron** » Mon Sep 06, 2004 6:12 pm

96Khz samples are useful if you want to play them back at different (i.e., lower) pitches. So you can do groovy pitch swoops with less obvious aliasing artefacts etc. Other than that the sample rates don't bother me (I kinda like my pitch swoops to have aliasing artefacts anyway

) but working at 24 bit is something I value, if only because, as somebody has already said, the noise floor can become apparent in certain processing situations.

Who makes the best 24bit 96kHz samples...