sample rates

[toddd]

ok, i know this topic comes up occasionally around here - the last time i recall it most people said they never record above 48k, for whatever reason.

but, i have been sending things in at 24/96 ever since i tried it for the first time and the drums i was recording seemed to come out of the mist at me (i know sample rate isn't the last word on clarity of recording, but for me i have always liked the higher rate).

now - i started talking to a person a couple of nights ago who is a drummer/engineer in a band who has done pretty well in the past (and continues to) and he said the words "empty bits" to me in regrds to my assertation of higher sample rates sounding better - he was sure 48k is fine ('all the big guys do it' i think was uttered). i have never heard anyone say "empty bits" before.

i was wondering people's opinions on this. in the age of dvd, i have always thought 96k would be good for posterity if nothing else: upsampling is a waste of time and hard drive storage is very cheap and only getting cheaper - so why not 96? (i guess sample rate conversions aren't the greatest thing in the world, but are they destructive enough to mess up the sound of a cd, when we're already ditherng anyway?)

and hasn't bob katz always said record at the highest resolution possible?

cheers, todd.

[jkhuri44]

gday todd..

let me start off by saying i have no where near the amount of experience as other people on this forum,

but i will say that for my latest project, i have recorded absolutely EVERYTHING in 24 96....and ....it sounds....great! and when i compare it to projects i recorded in 16 44 (bad bad idea) or 24 44....they just dont cut it!!!

some people have told me it depends what type of music theyre recording...for rock and metal and such, some people think the lower sampling rates work better, sound crunchier...and what not...

also, if your mixing ITB...higher sampling rates work in favour of plugins, especially compression...when your applying deeper compression, it tends to sound better with higher sample rates...as quiter stuff will be more detailed, and therefore less noisy when its cranked up....


and one last thing,

the cumulative effect of "noise" or crunchiness over tracks is an issue if you compare a 24 96 project to a 24 44/48 one...in my opinion....

just a few ideas.

[Howard Jones]

"Empty bits", eh?

This guy is working hard to maintain the reputation of drummers everywhere...

[otto ruiter]

The Neil Young CD I recently bought - "Live at Massey Hall 1971" comes with a DVD-Video disc. It has a 24bit 96kHz audio track and it sounds great!

[chris p]

"Empty Bits" is about bit depth (16, 20, 24) rather than sample rate (44.1, 48. 96, etc), and is to do with the ability of digital to represent quite parts with accuracy. In horribly simplisitic summary, you get one bit less in resolution for each 3db drop below 0dbFS, and so if you record at -24dbFS, that's the same as recording in 16 bit - all the higher end bits are "empty". So your drummer friend knows enough to say some words that he might not understand. Dithering is about truncating samples from more bits to less (typically 24 down to 16 for CDs) and again has nothing to do with sample rate.

The debate around sample rate (or at least as I understand it) is primarily one of utility. Originally, it was on the basis of 44.1KHz capturing waves up to 20KHz which is the threshold of hearing - no need for anything higher than that. Modern psychoaccoustics says otherwise (we may not hear 25KHz as a straight sine, but we can hear its modulations on lower frequency waves as timbre), so there's one reason to go higher, maybe. On the other hand, sample rate conversion especially non-integer division) introduces distortion, so if your target market is CDs, why introduce an extra conversion step that may distort the end product? The cynics go further and say if everything is going to be played as MP3 on iBud headphones, why worry about fidelity at all? (the less extreme argument is that your target audience does not have the gear to reproduce a 96Khz sampled recording with accuracy, so its wasted effort).

FWIW, I would record at 24 bits and whatever frequency from 44.1 up that you feel comfortable doing. At the end of the day, its more about the performance than the sample rate. Worry more about getting the right mics in the right position to capture that performance. Happy tracking.

[mfdu]

everything i do is at 24/96.

"lower sample rates sound more rock" ?? well, that might be tying into the old tape discussion of "ampex 456 has a smoother, more rock distortion curve, whereas GP9 is cleaner all the way to a snap into harsh distortion" . . .

with hard-drives as cheap as they are, i reckon that if i can fool myself into believing 24/96 sounds better, then why not use it?

chris
mfdu

[Kris]

For me - CD 44.1khz - DVD 48khz.

Some of my favourite albums were recorded on Mitsubishi X850 or Sony 3348 machines, all of them 16 bit. They still sound amazing today so I figure it can't be all that bad. Not a fan of 96khz myself. Tried it a few times, just not for me.

As they say... "if it ain't broke...."

[Kris]

http://cgi.ebay.com/Mitsubishi-32-Track ... dZViewItem

if only it was local and didn't weigh 300 pounds. You'd be looking at my new 32 channel A/D converter.

[Chris H]

Couldn't wait for 96k to become available and when i first used it was hapy with the improvement on accoustic stereo recording but thats all i considered it worth using for. Perhaps it was because back then the software didn't cope well with it. Must try it again on a project. I'm about to get an Ampex ATR 1/4" machine to mix to so will be ok to go high res through to final mastering.

[jkhuri44]

also, one thing i forgot to mention...

when using PT stock converters the 192 I/O....its an "ok" convertor...i find that the sample rate improves sound drastically...


but other people using higher quality converters, like for example Apogee stuff are ok with 44kHz sample rates because the convertor is more accurate, in the way it "resolves...bla blah blah (cant remember the technical argument)"..hopefully someone with more tech experience can finish that sentence off :P

[mfdu]

of course, if i could afford to do it i would track to tape each and every time.

that'd mitigate the question, to some extent.

and until i actually have some reasonable ad convertors, i'll stick to what i have convinced myself sounds best.

did i phrase that correctly?

regards,
chris
mfdu

[jkhuri44]

yeah, thats one way to say it :P


i was hoping for a nerdy explanation though! (insert Simpsons mathematician bloke saying something)

hahahahaha

[mfdu]

i could put my propeller hat on, but we'd be best to wait for howard to weigh in.

after all, the only thing my propeller hat achieves is a dramatic reduction of friends.

we're mostly agreed on 24bit, though, huh? apart from some, who aren't. gee - thats definitive.

i'm sure we could be saying "clock jitter blah blah spacial imaging blah blah frequency response blah blah relative positioning within time and space blah blah string theory"

15 inches per second.

chris
mfdu

[Chris H]

How about using 24 bit 96 to record the pristine sounds of a Les Paul played through a.......pod......

[mfdu]

urgh. i think i need to go and wash.

pods are for alien birthing clinics. or call centres.

[Ausrock]

we're mostly agreed on 24bit, though, huh? apart from some, who aren't. gee - thats definitive.

Guess I fit into the "who aren't group".....and contentedly so.

15 inches per second.

Sounds fine to me :-)

[Chris H]

I always go for the higher sample rate at the beer and wine festival......does wonderes for the stereo immaging......

[jkhuri44]

im sure we could be saying "clock jitter blah blah spacial imaging blah blah frequency response blah blah relative positioning within time and space blah blah string theory"

that sounds about right..replace the blah blahs with some more propeller words..

and we'd be looking at say...16/17 inches per second...

LOL

[Mark Bassett]

I highly recommend reading all these articles...

http://www.moultonlabs.com/main/cat/Hearing/

"Subjective expressions of audibility are often magnified greatly and non-linearly, as in when we say “The difference in clarity is amazingâ€

[NYMo]

hi there,

I'll throw my 2 cents in here....

(as i understand it) It is more destructive soundwise to do a sample rate conversion say from 48k to 44.1k than record at 24/44.1k in the first place.

I've never recorded any higher than 44.1.

When we are writing to standard 96k DVD's .i will then record at 96k ;-)

cheers
N
Y
M
o

Who has a 16 bit Sony 3348 ! (Kris ..do you need one?)

[mal stanley]

From my experience..if you are staying digital all the way( which is normally my workflow)I have always tried to stay 24/44.1 to avoid sample rate conversion..though some of the recent technologies have really improved...some plug ins especially compressors I understand can be more transparent at higher rates..some mastering engineers upsample for this reason....
The real benefit I believe of higher rates is to move the Nyquist cut off to higher up so there is less potential for artifacts due to the steep filtering necessary at 22.05 Hz...
There have been some fine recordings at 16/44.1 made over the years..the leap to 24 bit has brought the biggest difference in terms of detail and dimensionality as compared to analogue..
THe biggest difference you can make of course is good mics...good recording techniques..and good music....

[mfdu]

"...and good music...."

ok. somebody just had to pull that one out, didn't they.

you could record onto the side of a tin can with a sharp piece of wire, but if the tune has a good beat the kids will still jump to it.

chris
mfdu

[Kris]

Hey Nymo....

Do I need a 3348? Absolutely not. But I have been wanting a Mitsubishi X850 for about 10 years. I want to run in in record ready and use it as my A/D converter. And besides, it just looks so cool. Given that they were about $150,000us when new, the $875 the guy is asking for one now on ebay is really making me wish I lived in LA.

There's a 24 track version of the 3348 at Mixmasters if anyone is interested.

[mfdu]

i'm interested.

but i'd be happy with 24 tracks of analogue, and a client base who can afford it.

yeah ok, i should just take the expense of tape on myself. it's what any business should do, to achieve decent market positioning.
but try telling the wife that i'm now going to lose money on each session because i'm steadfastly pushing the tape barrow.

le sigh.

[NYMo]

Hi Kris,

I think you'll find the 3348 was nearly twice as much dollar wise as a X850 and should sound just as good..unless the 850 had the Apogee mod.

Plus..i might be asking a really reasonable price ;-)

Cheers
N
Y
M
O

[Kris]

Thanks Nymo.... not really in a position to buy right now. Still a future dream. Interestingly though, the X850 in question does have the apogee mod and also does 48khz.

[smash]

(as i understand it) It is more destructive soundwise to do a sample rate conversion say from 48k to 44.1k than record at 24/44.1k in the first place.

I've never recorded any higher than 44.1.

See that depends...

If you are say... sending to Rick to master... He will output out of digital system(A) at, for example, 96 kHz then send the signal through all his analogue goodness for it to be finally resampled on digital system(B) at whatever you need it to be... 44.1 for CD, 48 for Video, 96 for posterity, tape for kicks.

There is no "Sample rate conversion" happening at all.

PS - I'm not Rick, so I can't speak for him - but this is my recollection of the last project that he mastered for me.

[smash]

24 bit vs 16 bit

Putting a plugin onto a file that is 16 bit means that you end up with a resultant file that has a less than 16 bit resolution, due to the maths involved.

Putting the same plugin into a file that is 24 bit means that the resultant file when a plugin is applied *should* be undetectable as far as loss of resolution goes.

Also - 24 bit gives you a bit of a safety net when it comes to recording - you can leave a margin between 0dBFs and your actual recording level, without an audible difference, however when recording at 16 bit, for every dB away from 0dBFS you are there is a loss of resolution, but as soon as you go over you get evil digital recording.
24 bit makes it easier to get optimum level.

That said - there is definitely an audible difference between 16 and 24 bit - especially in the low level details like 'verb tails, but I doubt that any of us could hear the difference between a 20-bit and a 24 bit master, without a really, really nice monitoring environment.

[rick]

i bunch of people have been telling lately that 96 k on protools sounds scrappy
and one guy told me it was amazing when hd first arrived but after being converted he has now
reverted .

if your worrying about the sample rate its best play about untill you get satisfied that your not really sure
imo those theoretical rate numbers are all marketing horseshit in the washup

[David W]

Whilst i'm not sure what effects signals above 20K have on our perception of sound, or how the mathamatics are done calculating the pro tools mix buss one thing I know is that I will always pick what I think is the better option.

That to me is using best Mic Pre's, Mic's for the job, choosing (slowest speed (Highest Quality) when importing and converting files into pro tools, Recording and mixing at highest resolution possible to me (24/94K) thats at least what lavry convertors allow even though using PT 192.

Unless your short on prossessing power , Data storage or something of that nature i'm not really sure of why you would genuinely choose a lower resolution.

Another thing I am lead to beleive is that it has an effect on plug-in quality. Universal Audio have several on their UAD-1 cards that incorporate upsampling to 192K. I would hardly think that this would be a decision taken lightly given the extra prossesing power involved.

Whilst we can all argue black and blue whether our monitors, ears as well as our emotions allow us to hear these differences I will always take the higher resolution option, just in case someone else can here the difference.