Sampling rates in X1B?

First of all, it's been a while since X1b update was released, and I can say that NOW this is the product what I paid for. The X1 and X1a were...something else. ;O) I have succesfully started working with X1b full time now, with some great results! However, I'm still experiencing some odd behaviour. But not a single crash anymore! ;O)

The question is, what sampling rate would be the best to work with? Yes I know, this is almost impossible question, but I'd like to discuss from that side if someone else experiences these problems, and if they are my system or Sonar related.

My PC is Acer laptop, interface E-MU 0404 USB. So to begin with, 44.1 kHz seems to work well on most Sonar's own plugins (I mainly use X1's own plugs), no other odd behaviour, E-MU does not for some reason play any audio data in Windows 7 @ 44.1 when it is selected from E-MU control panel. So in Sonar the 44.1 works, but mostly I'm trobled by the thought that I could use better sampling rate.

48 kHz. This is for some reason the E-MU default setting, to which it sets itself if I play any audio data in Win 7, but still in reality the sampling rate IS 44.1! Works flawlessly in Sonar, no odd behaviour. Same thing with the idea of "low" sampling rate. 88.2 & 96 kHz both begin to produce very weird things, mostly with plugins, especially Studio Instruments, that string section plugin, which is very nice btw...if I do any ANY other control (than playing the actual keyboard), from MIDI-keyboard, such as bend, it creates some weird artifacts, like the key was stuck to play endlessly and the sound distorts. Using panic button.

In the manual CW says clearly that using 96 kHz is not recommended, because some plugins don't behave well at this sampling rate. Yeah, I've noticed. I'm okay with that, not that I like this fact. But then, WHAT IS a good performance/quality sampling rate in X1? Is there any clue to which sampling rate should I trust to get good results in quality, because I can already hear some serious artifacts caused by plugins in higher sampling rate. I would like to have some peace of mind, and I'm happy to stay in lower sampling rates if I know I'm having good results with it...

Anyone? Using latest drivers on interface and PC, check. Always using 24 bit resolution, check. I'm aware this may be mostly E-MU related problem, but I'm grateful of some other thoughts. E-MU itself seems to work well now on all sampling rates after the X1b update.

I'd say 48 Khz/24 bit is more than good enough for most. Most likely you wouldn't hear a difference between that and 44,1 Khz. Workflow- and dynamics-wise the bit depth is more important than the sampling rate. 24 bit gives you more headroom and your recordings are easier to control.

The problem with 96 Khz is that it often limits the functions of the soundcards. Many cards turn into one channel non-duplex cards with 96 Khz. Also, 96 Khz eats up a lot of HDD space.

bearing in mind you cant hear 30khz, but i doubt anything rock wise will have overtones that high, if you recording rock or something, 48khz is fine

The biggest differences in the recordings (we're not talking about heavy duty pro studios here) come from the microphones and room acoustics and the general recording skills of the person.
 
If you use 100 dollar mics in a not-so-well treated studio, the (possibly) higher audio quality also records the shortcomings of the gear and studio more audibly. An extra 100-150 dollar investment on a microphone surely improves the sound more than changing from 44,1 Khz to 88,2 KHz.
 
But whatever works for you, works for you.

Chregg


you have to look at the frequency content of wat you are recording, what i mean is if you think what you are recording has frequencies above say 30khz, then you'll need a sampling rate of over 60khz. the sampling rate has to be at least double the frequency of the highest frequency component in wat you are trying to record...google nyquists theorem

Actually you have to look at what we can hear and we can not hear 30Khz so there is no point in recording it.

UndertTow

Chregg


bearing in mind you cant hear 30khz, but i doubt anything rock wise will have overtones that high, if you recording rock or something, 48khz is fine

incase you didnt notice

My point is more fundamental and applies to all musical genres and even to any analogue to digital conversion: You only need to capture what you are going to use. This kind of approach is used in every field of engineering. It is only is the music world that we don't. That has more to do with lack of understanding combined with unscrupulous marketing and business than sound quality.

UnderTow

I dont think you can equate how far frequencies and harmonic information may extend according to a style or genre of music. All music is capable of creating energy far above our hearing range. Here is an example from an article in Sound on Sound (June 2002/ Synth Secrets) about how to synthesise a crash cymbal really well. So the humble crash cymbal may appear in any number of situations. Electronic music for example could feature a fundamental 4 Kz sawtooth wave and the first 10 harmonics of that would still have energy at 40 Khz.

But I also agree that you only have to go up as high as 48 Khz to do a good job.

A researcher at the California Institute of Technology, James Boyk, has measured the frequency spectrum of a crash cymbal, and found that, unlike that of other instruments, the cymbal's spectrum "shows no sign of running out of energy at 100kHz". He has also demonstrated that fully 40 percent of the sonic energy produced by a cymbal can lie above the commonly accepted threshold of human hearing at 20kHz. This raises two interesting questions. Firstly, if, as is generally accepted, most humans are unable to hear sounds above 20kHz, does the cymbal's supersonic energy still affect the audible sound in some way? Secondly, if it does, do we need synths, mixers, amplifiers and speakers with bandwidths exceeding 100kHz to reproduce these sounds accurately?
It's unlikely that anyone will be answering these questions at any time soon, but Boyk's research still has an important consequence when synthesizing cymbals: we must be able to create flat frequency spectra up to the accepted limit of human perception, if not beyond. This imposes constraints upon the type of synth we can use.

Thank you all for your thoughts. I've been studying this whole sampling rate subject quite a bit, through the eyes of electrician, have to mention, and there really does seem to be different opinions. Wiser ones than me can argue what sample rate truly is the best, but what I initially wanted to discuss about, was that at what sample rate X1 performs best? I mean, of course the system's performance is better with lower sample rate but is there any idea at what sample rate Sonar works best? At all rates, I suppose. That's what the salesman says at least. ;O)

And yes, my target is CD-quality. This is another thing, some say that using 48 kHz and then at the end converting this mastered masterpiece into 44.1 actually does more harm to the quality than using a target rate, 44.1. What do you think about that? I have a feeling that using 48 kHz gives more playground because it's more over the Nyquist. But these are feelings, yes, and damn do I still have some fantastic songs made with so bad quality I could cry...not my own ones. ;O) Music matters, of course, not the quality. But I would like to have such recordings I could use with confidence if I have a chance to work with professionals in high quality studio etc...

What do you think about the Sonar behaviour? Bad plug-in compatibility at those high sample rates?

Every plugin in the world is fully tested at 44.1Khz. Considering this and all the rest AND the fact it will put the minimum amount of load on your system, this makes it a no-brainer IMO.

UnderTow

if your happy with at 44.1 then fair do's mate, personally i like it at 48 k, then go to 16 when mastering !! preferences preferences lol.........cheers

Chregg

if your happy with at 44.1 then fair do's mate, personally i like it at 48 k, then go to 16 when mastering !! preferences preferences lol.........cheers

I can't detect any difference between 44.1 and 48 Khz with my converters and my monitors in a double blind test (nor can anyone else that has tried) but maybe you can with yours...

UnderTow

Primitive


I found some really interesting document, not sure if it has been linked in here already but anyway:

http://www.lavryengineeri...ts/Sampling_Theory.pdf

Very informative, but it still doesn't make clear statement. Maybe it's impossible, because this thing is too subjective. Or maybe the author didn't want to make clear statement.

Dan Lavry is very vocally against anything above 96Khz but says relatively little about 88.2 and 96 Khz themselves. Maybe because he himself sells products that are marketed as 96Khz or maybe because he realises that going against 96Khz is a losing battle. I don't know.

Most top converter designers I have spoken to or read about say that a sampling rate of around 60Khz is the maximum needed to cover all bases. That has to do with the anti-aliasing and anti-imaging filters used in the converters and not with the bandwidth needed to accurately digitise audible sounds. I think that with modern filter design even that isn't needed and 44.1/48 is enough.

Starting big tests with 44.1 vs 48 @ 24 bits, today.

This kind of test is harder than it might seem to pull of reliably. You need to record the same identical source at both sample rates simultaneously. (Different performances don't count). Then you need to convert one of the sampling rates to the other using a high quality Sample Rate Converter. I recommend SoX. And you then need to try to distinguish between the two versions in a double blind test. You can do this with the included ABXer addon with Foobar 2000 media player.

Just judging by listening to different recordings you have made will not give you accurate or reliable results.

UnderTow

The problem with Undertow is he can be extremely frustrating because he is pretty much always right. He's right again (IMO) on this topic: 48 khz for film, 44.1 for everything else.

On some converters it's WAY BEYOND expectation bias.

The difference between 44.1 and 96 is SHOCKING.

One thing to consider- does your recording interface have a zero latency feed back to your headphones? You won't be able to do as much at 88.2 or 96 as you would with 44.1 or 48 with the same WDM latency or ASIO buffer settings. Try setting your default new project setting to 96 kHz and create a new project and attempt to record a track. If you get weird clicks, pops, hesitance, etc go into Options->Audio and make your WDM latency or ASIO buffer size larger. This should smooth out playback.

The trade off for this is that it will make it harder to record if your interface doesn't have a zero latency feed back to your headphones. With zero latency you won't hear any FX you've placed on the track you are recording (so if recording with your guitar clean through DI and using guitar rig for your distortion/amp sound you would hear the clean tone only.) Everything will be in time, though. Without zero latency you will find it impossible to play in time above a certain WDM latency or ASIO buffer setting as there will be a delay between when you play and when you hear it in headphones.

To me the latency issue is much more important than the sample rate. As a general rule I record in 44.1 or 88.2 for audio only recordings and at 48 or 96 for audio that is going to be used with video. If you can do low or zero latency and still use enough plugins to mix down without crashing or running out of CPU or storage then I would try a higher sample rate. If not, 44.1 or 48 are fine.

Peace,

Tunes

In reading Bob Katz' Mastering Audio he had some really interesting thoughts on sample rates, clocking, etc. One of the things he mentioned is that higher sample rates aren't always better.

Essentially, 88.2 kHz recording on a jittery interface with not-so-great converters may not sound as good as 44.1 kHz on an interface with tight clocking and excellent converters. In addition, certain sample rates may sound better on certain devices than others because of the filtering used for each sample rate. Converters will use a filter to cut out sounds that are above what they can accurately capture, and some do this better than others.

The converters in your EMU are supposed to sound great as consumer recording equipment goes, so if you feel comfortable at 44.1 or 48 and don't have tons of horsepower/storage to spare, do it.