Fast PC, have to keep buffer at 1024?

  • Thread starter Thread starter mikel33
  • Start date Start date
Yeah I hear you man, but is that slight bit of detail worth/or causing your issues? From my understanding/experience, a great performance in a treated room with the right mic, can be done without need to worry about over sampling. You obviously posted because you are having an issue. Nobody here is judging you....

Without a sample of what you are asking about, it is just a bunch of guys guessing what the issue is. I can go down to 96 samples with my current interface at 30% CPU. But I am not using your interface.

I don't feel that anyone is 'telling' you what to do. Just asking questions to help resolve your issue. Obviously what you are doing now is not working for some reason. You likely either have a driver issue, or something that is not allowing your audio to stream well.


Try not to go on the defensive when someone asks you a question when trying to help you...
What do you mean defensive? Who's defensive? I'm not defensive. You're defensive!

Just kidding. :)

Regarding the technical issue, I learned about the dangers of rounding errors when I worked in aerospace in one of my other lifetimes. There we called them "tolerances" and getting them wrong was the difference between a rocket that flew and one that exploded. My introduction to audio recording was in the 80s and was strictly analog-based. I had (still have) a Fostex A8LR 1/4" 8-track and a Tascam MT-208 8 channel mixer. Then, my concern was mostly S/N ratio, which was a product of the "noisy-ness" of the mixer, the coercivity of the tape, the amount of magnetism of the tape heads and, of course, the narrowness of the tape tracks. I started doing digital video and learned about chroma sampling and the difference between a 4-4-4 space and a 4-2-2 one. In the digital video world, sample rate and bit depth effect, rather dramatically, video noise and digital artifacts. When I got into digital audio, I simply assumed that the same rationale applied and for the same reasons. Maybe it doesn't. I'm not an engineer and I'm certainly not a pro. However, as I said, it works for me. I have no issues to address in my recording. Though I'm sure my recordings could be improved, for me they're good enough. In fact, my biggest audio issues (aside from justifying my pricey toys to my wife) are Adobe's decision to move away from software ownership and figuring out Sibelius enough so that I can move away from Finale and use Sibelius' ReWire capability to record Sibelius scores in Sonar.

As for HR, I really like this site, which is why I've been hanging out here. I've learned a lot and, with the exception of a now-barred poster who wanted to teach the world how to do "industry recording" with $150 worth of hardware and pirated software, I like the people here -- they're smart, creative, knowledgeable and have a good sense of humor (and I like reading the News section, too). With that said, I have noticed a certain amount of "genre blindness" here -- there is an assumption that "music" means, "that which bands play in clubs and can be distributed on CDs or iTunes." Though a bias in favor of popular music is understandable, there are other genres that people write and record, and other reasons besides performing musician aspirations for writing and recording it. Here, I'll prove it:

Okay, everyone who writes and records classical music, raise your hands.

See, not only did no one raise their hand, but a bunch of people just groaned and left the room. I don't write classical, either, but I do compose, arrange and record music that is intended to be performed by "traditional" symphonic ensembles. Of course, that has nothing to do with bit depth or sample rate, but it is certainly relevant in a thread that, for example, asks, "What track do you record first?" I found it interesting how many people thought the drum track should be recorded first as that assumes that every recorded piece has a drum track.

Maybe the site needs an, "I'm not defensive" emoticon?
 
Actually man, there are a few like yourself around here. And I listen to 'everything'.

Even the crap that $150 creates. Funny enough, he just emailed me calling me a bitch. I could make his life miserable by intruding the dating site he is a member of. But I won't. He is 17 and just learning what is right and wrong. Not my place to judge anyone.

You are an asset to this site. I meant no disrespect in anything I have said. Keep in mind, we all have similar goals, yet many different genres and approaches. Take what advice from others is worth. It may not be right for your situation, but at least you have honest opinions from people with different situations. Whatever helps you in your particular situation, is for you to decide.

We all learn from each other. :)
 
Actually man, there are a few like yourself around here. And I listen to 'everything'.
I listen to ALMOST everything. :)

Even the crap that $150 creates.
I listened to the mixes on his website when he first posted. Odd thing was, I thought they were pretty good (not that I'm a particularly good judge of that kind of music) and told him so in a post.

Funny enough, he just emailed me calling me a bitch. I could make his life miserable by intruding the dating site he is a member of.
I'm going to guess that, with his attitude, he's not going to have much luck there, either. :)

But I won't. He is 17 and just learning what is right and wrong. Not my place to judge anyone.
Eh, I'm old -- I'll judge. Hey, you kids get off my lawn! ;)

You are an asset to this site. I meant no disrespect in anything I have said.
Thanks, I appreciate that. And I never took any of this as disrespect. One of the negatives of any website is that written words don't really convey the tone in which they were written -- I guess that's why we write music.

Keep in mind, we all have similar goals, yet many different genres and approaches. Take what advice from others is worth. It may not be right for your situation, but at least you have honest opinions from people with different situations. Whatever helps you in your particular situation, is for you to decide.
Noted. I guess I've read too many threads that begin, "What's the best inexpensive DAW for my PC?" and then someone posts, "Hey, man, you should get a Mac!"

We all learn from each other. :)
Yep. As I said, I've learned a lot here, and look forward to learning more. You've got a great website going.
 
My stock reply to AI/driver/daw/clickylatency issues...

Un install EVERYTHING! Drivers, Cubase etc.

Run a registry cleaner I have never had a problem with Ccleaner.

Disable On Board Sound. In BIOS if you dare. In Dev, Mang' if not

Disable Windows sounds, i.e. bleeps,bloops and whoops.

Reboot and run Ccleaner again.

Re install the fast track pro and then Cubase. Unfortunately my FTTP is in France with son but I am sure we used it at about 256 samples without issues. But why run at 96kHz? The received wisdom is that there is no sonic benefit and that many converters are in fact optimized for 44.1/48kHz and actually perform WORSE at higher sample rates. So you are probably doubling your CPU demand for no good purpose.

All that said the FTTP IS getting on a bit (I have also read that some M-Audio products do not work well with certain MOBOs running W7?) Retire it and get an NI KA6!

Ah! Just caught up with some of the other parts of this thread! Nobody is recording at 32bits! The FTTP will record at 24bits (and that gives you a noise level below Johnson noise for anything!) The internal processes in the PC could well run at 32 bits.

But the argument has raged for ever and no one has ever demonstrated, AFAIK that even 16bits at 44.1kHz can reliably be told from 24/96!
Dave.
 
Sounds like you got your head wrapped around the "mechanics, my personal LOL" of the issue as to 24/32 bit depth stuff. Could it be the LAN conx? Over my head. When I had a MOTU interface I could not get the damn thing working properly. The company was kind enough to replace it twice! Bad chip inside. Please do not be offended by my ? about your sample rate; if it works for you then OK. but high sample rates and short buffers have caused me trouble in the past. My rig direct monitors (in software) and I adjust buffer size if I need to on playback. I frequently mix 60+ track projects . Hence my recommendation to stem. It bums me out to read that your system is not doing what you want it to do. Wish I could help further. This is a great forum so keep coming back.
 
I had similar problem with an M-Audio Fast Track device. I believe they are junk. Try another AI.

They are most emphatically not junk.

I have used a ftrack pro on this old XP P4, a W7/64 HP desktop, W7/64 HP laptop, a WMCEd P4, (not supposed to work!) a Vista laptop, and a old P4 lappy with an 850mHz processor and 1/2 gig of memory. In each case the unit did everything that could reasonably be asked of it.

The plain fact is that not every audio device be it PCI, usb or fussywire will work reliably on every MOBO/chipset/os and some FAR more prestigious names than M-Audio have fallen foul of THAT fact!

Dave.
 
Just wanted to say I thoroughly enjoyed reading this thread and learned a bit about bit depths as well as PC nerd stuff, which is great because I just built a screamer a few months ago and am always looking for ways to improve it.

That being said, I remember reading something about the kid who said he did "industry recordings" for $150...does anyone know where I can find that thread? I'm feeling a bit bored and would like a good laugh.

Thanks for the great thread.
 
Sorry dude but that makes no sense at all.
Well, then let me explain:

The larger the ASIO buffer, the more time it takes for the computer to read it. However, if the CPU, memory and bus are fast enough, the amount of time it takes to read it will still be less than that which results in perceivable latency. The operative word, here, is "perceivable." For audio, the human ear will will perceive a delay when the latency is 10 milliseconds or more. If the computer can process a small buffer in .5 millseconds, and a larger buffer 1 millisecond, the perceived result will be the same, i.e. no perceivable latency.

My mixing computer is fast enough that, even with a relatively large buffer, e.g. 4k, the buffer data can be read in under 10 milliseconds, resulting in no perceivable latency.
 
Well, then let me explain:

The larger the ASIO buffer, the more time it takes for the computer to read it. However, if the CPU, memory and bus are fast enough, the amount of time it takes to read it will still be less than that which results in perceivable latency. The operative word, here, is "perceivable." For audio, the human ear will will perceive a delay when the latency is 10 milliseconds or more. If the computer can process a small buffer in .5 millseconds, and a larger buffer 1 millisecond, the perceived result will be the same, i.e. no perceivable latency.

My mixing computer is fast enough that, even with a relatively large buffer, e.g. 4k, the buffer data can be read in under 10 milliseconds, resulting in no perceivable latency.

As far as I understand it, latency is determined by buffer size, not the speed of the computer, although a faster computer may allow you to use a smaller buffer. Buffer size is measured in samples; the rate these are read is determined by the sample rate in use.
 
As far as I understand it, latency is determined by buffer size, not the speed of the computer, although a faster computer may allow you to use a smaller buffer. Buffer size is measured in samples; the rate these are read is determined by the sample rate in use.

tHat.^. is also my understanding....Found this...

uffer size/latency:

At 44.1 kHz, a buffer size of 256 samples results in a 5.8 millisecond delay, 128 samples in 2.9 ms delay, etc. (halve or double the number)

Anything above 256 samples is usually not acceptable while recording/monitoring directly, while it might be acceptable while mixing or producing. Some sound cards allow for zero latency monitoring via hardware routing, such as RME.

The total latency is somewhat higher though, as you need to take converter latency (e.g. 2 x 0.5ms) and other processing latency into account. Once the total latency is above 11 ms the human ear (or rather: brain) starts to take notice. Most trained musicians will start noticing a lot earlier than the average joe.

32 samples is 0.7 miliseconds
64 samples is 1.5 miliseconds
128 samples is 2.9 miliseconds
256 samples is 5.8 miliseconds
512 samples is 11.6 miliseconds
1024 samples is 23.8 miliseconds

Approx.

That is, for a given sample rate the delay is a function of sample number.

Dave.
 
32 samples is 0.7 miliseconds
64 samples is 1.5 miliseconds
128 samples is 2.9 miliseconds
256 samples is 5.8 miliseconds
512 samples is 11.6 miliseconds
1024 samples is 23.8 miliseconds

I found those numbers interesting. and I thought I'd see what numbers my Firestudio Project spits out. Using Universal Control to set the sample rate, and Studio One to see my latency, I get these:

64 - input: 1.90ms/84 samples > output: 3.90ms/172 samples
128 - input: 3.36ms/144 samples > output: 5.35ms/236 samples
256 - input: 6.26ms/276 samples > output: 8.25ms/354 samples
512 - input: 12.06ms/532 samples > output: 14.06ms/620 samples
1024 - input: 23.67ms/1044 samples > output: 25.67ms/1132 samples
2048 - input: 46.89ms/2068 samples > output: 48.89ms/2156 samples
4096 - input: 93.33ms/4116 samples > output: 95.33ms/4204 samples

Now, I usually have it set on 256. This seems to give what seems to be a total of 8.25ms, and only with a consistent 2ms delay between input and output, no matter the sample rate. I assume this is my ideal rate. But would there be any real advantage to going down to 128, or even 64? I'm going to guess that it wouldn't really be a big deal in this case, and probably continue using 256. Because it works fine, and hasn't made anything difficult. I just ask out of interest.
 
I had a M Audio Fast Track as well that had the same issues, I tried everything, finally I gave up and bought a Akai EIE Pro and everything was fine. Must be something screwy with the drivers, I was using XP at the time and both my Daws had the same issue.
 
Back
Top