R/E/P Community

Please login or register.

Login with username, password and session length
Advanced search  

Pages: 1 2 [All]   Go Down

Author Topic: sample rate debate  (Read 6157 times)

Adam Miller

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 255
sample rate debate
« on: December 08, 2009, 07:59:55 am »

Now... I find talking about sample rates as tedious as the next man... I'm just slightly intrigued as to why most people seem to working at 48k. What are you gaining?

Ad
Logged

Miguel M.

  • Newbie
  • *
  • Offline Offline
  • Posts: 16
Re: sample rate debate
« Reply #1 on: December 08, 2009, 08:49:37 am »

I notice an improvement while working at 48 kHz mainly because some plug-ins sound better at this sample rate. They also sound better at 96 kHz but that's too much disk space and cpu usage.

And, i'm not 100% sure about this so please someone chime in and confirm it (or not), but i've read that most AD and DA on the market are internally designed to work at 48kHz and it's multiples values (96 kHz, 192kHz).
Logged

j.hall

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 3787
Re: sample rate debate
« Reply #2 on: December 08, 2009, 03:54:43 pm »

Adam Miller wrote on Tue, 08 December 2009 06:59

Now... I find talking about sample rates as tedious as the next man... I'm just slightly intrigued as to why most people seem to working at 48k. What are you gaining?

Ad


more resolution.
Logged

Rob Darling

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 294
Re: sample rate debate
« Reply #3 on: December 08, 2009, 04:29:45 pm »

There is no more "resolution" when increasing  the sample rate of a recording, only increased bandwidth.

The difference between 44.1 and 48k is inaudible in a well-designed convertor.  It will shift artifacts only slightly.  You will be increasing the max frequency recorded from 22k to 24k, only a small part of an octave at this frequency.  

That said, in lesser devices with heavy artifacting, shifting artifacts a little MAY be audible since different frequencies will be aliased and will re-enter at different places, though I haven't found a piece of gear on which it was audible in a really, really long time.

I offer this story as an example which illuminates the value of increasing sample rate:

I once was called in to put together a country studio for one of my clients.  He was doing a new record that would be very, very acoustic in what was to be a gorgeous room and was curious about doing it at 88.2.

I told him it definitely made a difference as the tracks stacked up and as you began applying eq and dynamics, but that it would be a major commitment since this was before PT HD and it would require using Nuendo and host-based recording.

Armed with this info, he went to his producer and engineer on the record, a team that had worked together for 20+ years- on the same records, made in the rooms, with the same gear, on the same monitoring, every day for a couple decades.

They had two totally different responses:

One said absolutely, it made a difference, the other said no way, it wasn't worth it, hogwash.

I was a little mystified, because in every way, these guys hear the same and think with one brain.

When I began talking to them, I found the difference:

One had evaluated high-res on super-badass convertors, on two-track material.  He had found only subtle improvement.

The other had evaluated high-res on some fairly cheap stuff, transferring multi-track and trying to mix.  He'd found a huge difference between working with the 44.1 and 88.2 transfers.

So the ideas are:

Nyquist being what it is, phase shift and aliasing are necessary aspects of digital recording.

Recording at increased sample rates pushes the artifacting of these phenomenons out-of-band, above where we hear.

Cheap convertors benefit greatly by this.

Great convertors benefit less.

As you get more tracks and manipulate them, the benefits become more apparent- as you eq and compress, you work against the artifacts at 1x, but at 2x or higher, the artifacts will be out of band.  It's easier to get a smoother, more "analog" mix when working at high sample rates.  EQ and compression take hold much more naturally without feeling like they add so much bite.

YMMV, but working at a 48 over 44.1 won't sound better, though it might sound a little different on lesser gear.  You will still have in-band aliasing and phase shift, it will just be different since aliasing is not musical (think ring-modulator) and will give different artifacts.

Working at 48k, in general, for music, is a bad idea.  It CONSTANTLY leads to confusion and problems since most people by default work at 44.1... witness the number of mixes that still come in on IMP at the wrong speed:)  

And while shifting from 44.1 to 48 for your samplerate in record/mix will only make your aliasing different, applying SRC is truly destructive- it is just adding another eq to everything.

-rob darling


Logged
____________________
rob darling
rob@robdarling.net
www.robdarling.net

Josh McArdle

  • Newbie
  • *
  • Offline Offline
  • Posts: 42
Re: sample rate debate
« Reply #4 on: December 08, 2009, 04:59:47 pm »

I work at 1million bit/26000 kHz and I can definitely hear the difference through my Behringer Eurodesk.

...

Rob, thanks for the explanation. Personally I couldn't tell the difference between 44.1 and 48, but I don't have uber-ears yet...FWIW I tend to stick with 44.1 for most things, just makes life slightly easier when the band wants a CD at 4 o'clock in the morning Smile

I haven't yet experimented with anything above 48k, but this all makes me very interested to try it out...
Logged

jetbase

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 2454
Re: sample rate debate
« Reply #5 on: December 08, 2009, 05:51:39 pm »

I use Prism converters for ADC & an Alesis HD24 for DAC (as well as additional ADC). The HD24's clock is much more stable at 48kHz than it is at 44.1kHz. Even when clocked off the Prism everything sounds better, IMO, at 48kHz in my setup, including after final SRC to 44.1kHz.
Logged
sleep is not an option

jwhynot: "There's a difference between thinking or acting dogmatically and drawing from experience."


Glenn Santry
http://www.myspace.com/glennsantry

Seb Riou

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 284
Re: sample rate debate
« Reply #6 on: December 09, 2009, 02:17:59 pm »

Anti aliasing filter set 2kHz further.
As it may be the weakest point in the converter, the further the better for me.
Logged

KB_S1

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 931
Re: sample rate debate
« Reply #7 on: December 09, 2009, 04:43:10 pm »

Seb RIOU wrote on Wed, 09 December 2009 19:17

Anti aliasing filter set 2kHz further.
As it may be the weakest point in the converter, the further the better for me.



This is what I have heard from most digital design guys, including the Prism man.
Unfortunately CD is stuck with its format and filters I believe?
Logged
<a href="http://www.parklanerecordingstudios.com/" class="link3">Park Lane Studio</a> Where to find me most of the time<br /><br />

<a href="http://www.flickr.com/photos/kb_s1/" class="link3">Flickr</a>where to see what I have been up to  <br /><br />

Seb Riou

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 284
Re: sample rate debate
« Reply #8 on: December 10, 2009, 02:29:12 am »

Yes, but chances are your mastering engineer (hence the use of a mastering studio) have superior quality converters with superior analog stage and superior filter quality, so I'd leave him/her with the tricky reduction to 44,1 (not even speaking of dithering to 16 bits)
Logged

Rob Darling

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 294
Re: sample rate debate
« Reply #9 on: December 10, 2009, 05:39:23 am »

Seb RIOU wrote on Wed, 09 December 2009 19:17

Anti aliasing filter set 2kHz further. the further the better for me.


Moving the passband 2k higher does not move the artifacting out of band in the case of 48k- it just makes it different from 44.1.

All of the phase shift and aliasing of 48k will still be in the audible area.

It will be different frequencies being aliased and they will be introduced at different frequencies from 44.1, but there will be just as much aliasing at 48k as there is at 44.1.

Moving to 88.2 will move aliasing and phase shift out of your audible range.

Moving to 48k will not.

Moving to 48k may be something that sounds different from 44.1, but any preference you have is only subjective.  You have not changed any amount of aliasing or phase shift in changing from 44.1 to 48.

Working at 48k for music when the final destination is 44.1 and CD is a bad idea.  

It often creates confusion.

It will require an additional destructive step of sample rate converting from 48 to 44.1 and fixes nothing, it only changes the tone of the problem.

-rob darling


Logged
____________________
rob darling
rob@robdarling.net
www.robdarling.net

Seb Riou

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 284
Re: sample rate debate
« Reply #10 on: December 10, 2009, 07:03:02 am »

That's why I record and mix at 88,2

but if I HAVE to work at lower rate, I'd take 48kHz

Call me stubborn
Logged

NelsonL

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 1233
Re: sample rate debate
« Reply #11 on: December 10, 2009, 08:18:45 am »

IMP sample rate errors are funny, but if your ME pulls that on you... run.

Correct me if I'm wrong, but SRC seems moot if your mastering with an analog chain or printing your mixes to tape, as I prefer to do.

In either of those cases, your ME may or may not SRC, depending on their work flow. In every case, they'll do what they think sounds the best-- as should we all.
Logged

grantis

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 1407
Re: sample rate debate
« Reply #12 on: December 11, 2009, 11:21:36 am »

I understand the "no debate-what sample rate" thread is noise free....so....

I'm somewhat surprised to see how many people are working at 24/88.2.  Personally, I believe it does sound slightly more "open" and the picture is a bit more "complete" at 88.2.

Unfortunately....my handy editing tool...the Mbox Micro.... only runs at 48k.  Otherwise I'd be running at 88.2 as well.

I'm curious to know....of those folks who are running 88.2, what hardware are you using?  PTHD?  PTLE with a monster machine?  Something else?

Logged
Grant Craig
Nuovo Music (Me)
Skiddco Music (Where I work)
Work History (Well, some of it anyway)

Seb Riou

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 284
Re: sample rate debate
« Reply #13 on: December 12, 2009, 02:05:41 pm »

RME fireface and two ADI-8 QS
Logic 8
Logged

Nick Sevilla

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 853
Re: sample rate debate
« Reply #14 on: December 14, 2009, 02:12:06 pm »

Thanks for all the great info.

On my next album I will record at 88.2 KHz / 24 bit. I do have 2 internal 15,000 rpm 320 GB Raptor hard drives on my Mac Pro, so I should be fine with hard disk bandwidth.

And it will probably be an all acoustic album in which I'll push the system not very much, by utilizing microphones and outboard to get the bulk of the sound, and mixing with less plug-ins afterwards.

The Final Masters I will keep at 88.2KHz / 24 bit and allow the Mastering Engineer the final sample rate and bit rate reduction.

Cheers
Logged
-------------------------------------------------
It is quite possible, captain, that they find us grotesque and ugly and many people fear beings different from themselves.

www.nicksevilla.com

Rob Darling

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 294
Re: sample rate debate
« Reply #15 on: December 18, 2009, 08:43:58 am »

NelsonL wrote on Thu, 10 December 2009 13:18

IMP sample rate errors are funny, but if your ME pulls that on you... run.

Correct me if I'm wrong, but SRC seems moot if your mastering with an analog chain or printing your mixes to tape, as I prefer to do.

In either of those cases, your ME may or may not SRC, depending on their work flow. In every case, they'll do what they think sounds the best-- as should we all.


Mixing ITB is very common for a lot of people, in particular around here.

Creating a multitrack at 48 is not an ideal thing to do if it is not destined for analog.

If you have complete control of a project and it is your own, do as you will.

But as an engineer for hire, creating a 48k project for a client is not a such a good idea.

r
Logged
____________________
rob darling
rob@robdarling.net
www.robdarling.net

NelsonL

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 1233
Re: sample rate debate
« Reply #16 on: December 18, 2009, 09:56:03 am »

robdarling@mail.com wrote on Fri, 18 December 2009 05:43

NelsonL wrote on Thu, 10 December 2009 13:18

IMP sample rate errors are funny, but if your ME pulls that on you... run.

Correct me if I'm wrong, but SRC seems moot if your mastering with an analog chain or printing your mixes to tape, as I prefer to do.

In either of those cases, your ME may or may not SRC, depending on their work flow. In every case, they'll do what they think sounds the best-- as should we all.


Mixing ITB is very common for a lot of people, in particular around here.

Creating a multitrack at 48 is not an ideal thing to do if it is not destined for analog.

If you have complete control of a project and it is your own, do as you will.

But as an engineer for hire, creating a 48k project for a client is not a such a good idea.

r



Mixing ITB is moot though, unless you're effectively mastering the thing yourself, there's no need to SRC anything other than a ref.

From another standpoint, given the amount of home recording going on these days, including projects that have professionally tracked basics, I can see how creating a session at a sample rate above 48k could be a bad idea. I'd rather work at 44.1 or 48 than have the masters SRC'd.

I can see your point Rob, but given the popularity of 48k sessions, I'd expect any engineer for hire to handle even an unlabeled session correctly. In fact, that's just one more way in which IMP is a great learning tool.
Logged

j.hall

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 3787
Re: sample rate debate
« Reply #17 on: December 18, 2009, 02:55:08 pm »

i guess i just don't see how it matters. 44.1 or 48.  why is one a good idea or bad idea?

88.2 and 96 sound better to me.  but i honestly just don't care.  if you send me something to work on, it is what it is, and i'll do the best job i can either way.
Logged

NelsonL

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 1233
Re: sample rate debate
« Reply #18 on: December 18, 2009, 04:14:22 pm »

Remember D-88s? I think it was more relevant in those days.
Logged

Rob Darling

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 294
Re: sample rate debate
« Reply #19 on: December 23, 2009, 07:24:41 am »

NelsonL wrote on Fri, 18 December 2009 14:56

robdarling@mail.com wrote on Fri, 18 December 2009 05:43

NelsonL wrote on Thu, 10 December 2009 13:18

IMP sample rate errors are funny, but if your ME pulls that on you... run.

Correct me if I'm wrong, but SRC seems moot if your mastering with an analog chain or printing your mixes to tape, as I prefer to do.

In either of those cases, your ME may or may not SRC, depending on their work flow. In every case, they'll do what they think sounds the best-- as should we all.


Mixing ITB is very common for a lot of people, in particular around here.

Creating a multitrack at 48 is not an ideal thing to do if it is not destined for analog.

If you have complete control of a project and it is your own, do as you will.

But as an engineer for hire, creating a 48k project for a client is not a such a good idea.

r



Mixing ITB is moot though, unless you're effectively mastering the thing yourself, there's no need to SRC anything other than a ref.

I can see your point Rob, but given the popularity of 48k sessions, I'd expect any engineer for hire to handle even an unlabeled session correctly. In fact, that's just one more way in which IMP is a great learning tool.



It's not moot.  Unless your ME has two separate sets of DA and two mastering rigs- one to play out of and one to record back into- they will be performing SRC.  This may likely be an eventuality.

As for counting on the intelligence of anyone other than yourself expecting other people to figure out that you are working outside of standard (and working at 48k for music is not standard), this is not a great practice.  

This is what standards are for- to make sure that when you share your work, it will have the maximum value.
Logged
____________________
rob darling
rob@robdarling.net
www.robdarling.net

Rob Darling

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 294
Re: sample rate debate
« Reply #20 on: December 23, 2009, 07:33:13 am »

j.hall wrote on Fri, 18 December 2009 19:55

i guess i just don't see how it matters. 44.1 or 48.  why is one a good idea or bad idea?

88.2 and 96 sound better to me.  but i honestly just don't care.  if you send me something to work on, it is what it is, and i'll do the best job i can either way.


If you send a 48k project to an ME they may likely need to SRC.  This will affect the sound.  If you work at 44.1, it will not require this.

If you work at 48, other people who inherit the project may not realize and it will be a problem.  You cannot count on the intelligence of other people.  

Working at 48 than 44.1 is not better in any case, it is just different, since it has just as much aliasing and phase shift, if you are even using gear that has enough problems for the difference to be audible.  So making the decision to create potential issues downstream is not a good idea for what is in fact just changing the tone of the problem, not fixing it.

As for working at higher sample rates- yes, it will make it much easier to mix digital source material by pushing two of the prime issues of digital recording- phase shift and aliasing with brickwall filters- out of the audible band.
Logged
____________________
rob darling
rob@robdarling.net
www.robdarling.net

Rob Darling

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 294
Re: sample rate debate
« Reply #21 on: December 23, 2009, 07:40:16 am »

NelsonL wrote on Fri, 18 December 2009 21:14

Remember D-88s? I think it was more relevant in those days.


Yes, early 90's prosumer digital sucked.  

r
Logged
____________________
rob darling
rob@robdarling.net
www.robdarling.net

NelsonL

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 1233
Re: sample rate debate
« Reply #22 on: December 23, 2009, 11:19:19 pm »

So Rob, if I understand you correctly, the professional standard (in your estimation, or is this an AES thing or what?) is that you can use any sample rate except 48k for music?

Here's a thread in which several regular PSW contributing MEs discuss running their "capture" AD/DA at 24/44.1 -- I think this is more common than your post would suggest. As you can see, it doesn't necessarily require a complete "second rig," although approaches differ:

    http://recforums.prosoundweb.com/index.php/mv/msg/29426/0/0/ 898/

Now, all that being said, I have to admit that the album I've been helping sequence this week is a mix of 44.1 and 88.2, and the difference in just the reference mp3 files is enough to get me thinking of moving back up again.
Logged
Pages: 1 2 [All]   Go Up