A theory about digital ticks and pops and solution

V

VTmosaic

New member
We have been chasing the oft-referred to pops and clicks, ticks, etc., for the past few weeks. We have taken pretty much all so-far suggested steps to optimize our machine and operating system, and they all cumulatively have improved things, one little bit at at time. I've learned a lot about keeping most of my tracks archived and not using effects while tracking, as well, which helps a great deal in the quality of our tracks.

This forum has been a HUGE help in finding these steps and making it possible for us to make progress. Thank you so much.

But there seem to be other causes, as well, that remain constant, and I am starting to form a hypothesis, and would really appreciate feedback from some audio-techie expert(s) who might have time to consider and discuss this question.

Is it possible that to some extent, those ticks that seem to remain despite all our computing power and optimization steps, and occur with maddening consistency when recording high frequency sounds are due to the fact that there simply aren't any frequencies available in that input signal that are much above the human's "stardardized" audible hearing range (20kHz or so, right?), and I'm using the 96kHz setting to record my audio?

I know that parapgrah was long and confusing. Let me try restating and also giving examples. If the source of the audio is not delivering any significant amplitude (and even perhaps totally cut, no sound at all) in the frequencies above the presumed standard of 20 kHz (for which 44.1kHz sampling rate is fine), but we're sampling at 96 Khz, is it possible that there is "empty space" at the highest frequencies captured by the 96kHz sample rate? Could this also account for noise that intrudes when we use a guitar effects box where the effects might have heavily attenuated the highest frequencies or even might not have any in that range?

Maybe this is a stupid newbie question, or maybe I'm onto something.

Here's what got me thinking this. We had 5 songs done for us for free by a nice guy with a nice new studio and DAW who needed guinea pigs. He did a nice job, but when he gave us the mixdown, all we got were the .cda files. For various reasons, I decided to record the 5 .cda songs off the CD into Cakewalk Home Studio XL2004 via our Delta 1010's monitor mixer (so I could put them on the same CD as the new stuff we've been recording without getting clipping).

One song has some very high frequency chimes in the beginning. It's a very quiet place in the song. And I got this AWFUL zipperlike ticking at the high point of every beautiful chime note! It was really bad. What fixed it: recording at 44.1 and 16 bits.

That got me thinking: Was Cake Walk and/or the Delta 1010 AD converter (as would any other recording system capable of greater word length as well as higher sampling rate) filling in the missing digits with noise? Could I fix this problem by recording these 5 songs at 16 bits and 44.1kHz sampling rate? I tried it AND IT WORKED!

And what about analog input devices like clunky old microphones that just can't get the higher frequencies? Or even digital devices, (like a guitar effects box that's got the higher frequencies squished right down or out) that are in the input chain during recording? It's not even clunky, it just isn't "optimized" for high fequency sampling.

Is this possible that these each get more noise at higher sampling rates and even word lengths because there's just nothing there for the A/D converter to work on? It's not usually very noisy, actually, but when you're really listening, it's definitely there. And much worse on high frequency sounds. Plus it can "build up" if enough tracks each have even a little bit of it.

I'm plan to experiment with this hypothesis as I work to improve the quality of the tracks we're laying down. I didn't find a lot about this topic in any of the knowledge bases in mine or my partner's research. That doesn't mean it wasn't discussed, just not in conjunction with the "ticks and pops" problem (which we saw EVERYWHERE).

I thought I might throw this out, though, to see if others have more information on this possibility, and can give some pointers an how/when to use highest quality recording capabilities and when to pull them back a bit to get better results?

Thanks in advance, and maybe I will get a chance to run my question in person past some folks this next weekend, at Jam Fest. I'm looking forward to this, but nearly as much as vtgreen81 is.
 
TripleJ has is right although the two items are intertwined. When recording at higher qualities (96k/24b) the files are well over twice the size of recording at 44.1k/16b. You could be having a processing problem, or being that the CD is playing audio from a file that is 44.1k there could be a compatability issue there.
You might find it much easier to find a CD ripping utility (there are plenty of shareware versions out there) and rip the CD audio to wav files. This would also save you one level of going from digital to analog to digital, which always creates noise.
-k
 
The basic problem is that you are trying to stream a large amount of information and not all computers are up to it.

Windows machines (no flame I use a P4) are a kludge of mismatched parts from many different manufacturers and not all are in tolerance with each other. Add the inconsistencies of various flavors of Windows and its updates to various manufacturer's drivers and programs installed and you have "little problems" to solve.

I was lucky. My current DAW is a pure Intel mobo with an Intel P4, high quality memory, Win2k SP4 and fast drives with a good stable (Echo Mia) soundcard. My first test was 35 stereo 24-bit/44.1khz tracks- 15% CPU useage and I have ZERO problems.
 
TripleJ said:
96khz is the frequency of sampling (sampling rate), not the frequency of sound.
Click to expand...

I know, but the two are related, as I understand it so far. Are you pretty well informed about the technology of digital recording? I'm dangerous, in that I have studied this stuff on an introductory level (we took a really good class that covered these topics at a intro level). So, I know a little, just enough to be trouble.

Sampling rate needs to be at least twice the highest possible frequency of the sound that will be recorded. The point of high sampling rates is to get as many samples of a single wave as possible in the time that wave is passing through the AD converter. Two few samples and the computer doesn't have enough data to really rebuild the true frequency dynamics. Too many and we overload the processor and storage of the computer with HUGE file sizes.

So, this has all come together in my head with my recent struggles with digital noise. And my discovery that I get digital noise in high frequency waves if I record from a CD ( reproduced using the 44.1kHz sample rate) to a project at 96kHz. The facts are: if I reduce sampling rate and bit rate... voila, noise is pretty muich gone!

Even if WE can't hear frequencies much above 20Khz (on average), the computer doesn't have that problem: as long as the recording input device is sensitive enough to pick up the sounds that are there but we can't hear, there is something there for it to convert.

But what happens when there is practically NO sound at the higest frequencies? what if it's sampling an already corrupted wave formed from a low sampling rate (playback from CD)? Or what if it's getting its feed from a guitar effects box with parameters in effect that completely cut the highest frequencies?

Then, does the Signal to Noise ratio comes into play? If the AD converter is getting NO sound pressure level readings at the higher frequencies, then would the SNR would be such that there would be some noise, and faithfully reproduce it?

Is that possibly where my stubborn (and only high-frequency wave-related) clicks and pops are coming from?
 
Actually - the remaining clicks/pops could be computer limits as Tim suggested, but there are also two other possibilities.......

1) A clocking issue - or 2) completely unrelated - you say you're only hearing "clicks" during moments of high-frequency signals? This suggests you're possibly recording at too-hot a level and signal transient peaks are causing clipping.

This is quite possible regardless of not seeing it on the meters, since very fast transients could still create an issue and yet never register visibily.

Also pickup a copy of the digital Bible - John Watkinson's The Art of Digital Audio.... digital audio doesn't work quite the way you seem to think it does! ;)
 
VTmosaic said:
But what happens when there is practically NO sound at the higest frequencies? what if it's sampling an already corrupted wave formed from a low sampling rate (playback from CD)? Or what if it's getting its feed from a guitar effects box with parameters in effect that completely cut the highest frequencies?

Then, does the Signal to Noise ratio comes into play? If the AD converter is getting NO sound pressure level readings at the higher frequencies, then would the SNR would be such that there would be some noise, and faithfully reproduce it?

Is that possibly where my stubborn (and only high-frequency wave-related) clicks and pops are coming from?
Click to expand...

No. If there was no audio at that frequency then there is nothing. Nothing does not equal pops and clicks. It is just nothing.

If I were you I would just work in 44.1khz and see if that works okay.
 
Blue Bear Sound said:
Actually - the remaining clicks/pops could be computer limits as Tim suggested, but there are also two other possibilities.......

1) A clocking issue - or 2) completely unrelated - you say you're only hearing "clicks" during moments of high-frequency signals? This suggests you're possibly recording at too-hot a level and signal transient peaks are causing clipping.

This is quite possible regardless of not seeing it on the meters, since very fast transients could still create an issue and yet never register visibily.

Also pickup a copy of the digital Bible - John Watkinson's The Art of Digital Audio.... digital audio doesn't work quite the way you seem to think it does! ;)
Click to expand...

We were playing with a cricket last night! It gave me great examples of my problem. When I recorded that cricket at 96kHz, 24 bits, every single high point of that little cricket's song when click. I reduced various settings and retried... and only when I got to 44.1/16 was the noise slightly decreased and at least random enough that I could mostly ignore it. It no longer clicked on every chirp. Exactly what happened when I was recording the chimes.

This seemed to be the case no matter how low the recording level. We had the mic through our Yamaha mixer, I never even got to the mid-point of the mixer's meter even at max gain. Is it possible that really high frequencies are loud enough to cause clipping but not even get the meter into the midrange?

Maybe it is just the computer really is not gonna work with its current config. Maybe we need to get a more run of the mill display adapter and forget running two monitors, and see if that helps. Maybe recording at 44.1/16 is reducing the load on the processor and we're getting slightly better results?

Or maybe it's Cakewalk HSXL2004? I can't compare Cakewalk to Audacity on this issue because Audacity can't do 96/24. It gets about the same amount of noise as HS at 44.1/16, I think.

Re: Clocking issue: I will follow up on that, as well. There are several different options on the Delta 1010 and I don't know a lot about it. Thanks!

Gotta run, so I can get 5 days work done in 3 so we can go do Jam Fest.

I will get that book. As well as review my own class notes and textbook from the class that gave me such crazy ideas. I have probably mushed some concepts together and then created a false model in my mind, like the three blind men describing an elephant! Hey, that's what learning is about. I appreciate the pointers.
 
Post the specks on you computer, that would help a lot. CPU, memory, hard drive(s), everything. We might see something.
BTW if you are running two monitors, you vid card might be a resource hog. We covered a program called "power strip" a while back that lets you de-prioratize your vid card, along with other unimportant resources, and make the CPU pay more attention to what you want it to. Helped me a lot with my FW 410.
-k
 
radzikk said:
TripleJ has is right although the two items are intertwined. When recording at higher qualities (96k/24b) the files are well over twice the size of recording at 44.1k/16b. You could be having a processing problem, or being that the CD is playing audio from a file that is 44.1k there could be a compatability issue there.
You might find it much easier to find a CD ripping utility (there are plenty of shareware versions out there) and rip the CD audio to wav files. This would also save you one level of going from digital to analog to digital, which always creates noise.
-k
Click to expand...

I know that there is more data to be processed and captured at 96kHz sampling rate, and that's why we built the most souped up box we could to process the signal from the Delta 1010.

But, sampling rate is also related to the frequencies to be recorded, I know for a fact, though I may have the concept twisted in my mind. I know we sample at 44.1 because generally humans can't hear over 20kHz frequencies. Here's where I"m probably screwing up the concept: I think that's so a 44.1kHhz sampling rate assures at least two points are captured on even the highest frequency sound we can hear. I have charts and diagrams that I will revisit and restudy (when I can get a breather, after Jam Fest, I guess).

Maybe that isn't related in the least with my problem, but I'm gonna wonder about it until I can fully understand how it isn't so. If you sample at too low a frequency, then the wave that is reproduced from those samples may be missing important points on a high frequency sound wave. If the sound wave is over and done with so fast that we get too few samples, and the highest highs are missed, then when happens when that sound wave is reproduced? Is it distortion or just less bright sound?

Re: ripping software, I tried that, believe me. I have no long-term need to spend money on a really good ripping program since we are only doing this for this special situation. I do use a slightly more powerful version of Real Audio (a premium version) and it does a fine job for commercial CD's when I want to do a compilation.

Not for this thing I was trying to do, though. These are our own songs recorded at someone else's home studio, given to us on CD in .cda files and whenever I tried to convert them from .cda to wav or any other format by whether I used Real Audio or even Windows or Winamp, they clipped like crazy. I could duplicate them CD to CD, I have a CD-ROM adriver nd a burner and just copied the files from one to the other, and then we had no problem, but I wanted to have them on the same CD as the songs we're now recording ourselves.

I decided I'd just record them off the CD into HomeStudio and then I could get rid of the parts that were causing clipping (used a volume envelope and quieted the few problem areas down). After that, I could convert them to .wav files and put them onto the same CD as the new stuff we're recording now.

In the process of doing that, I found this excellent example of the problem being only on high frequencies. But since then, we found out it's not because it was from CD. We had the same problem recording a chirping cricket, live.

Thanks for trying to help me get this straight in my mind!
 
radzikk said:
Post the specks on you computer, that would help a lot. CPU, memory, hard drive(s), everything. We might see something.
BTW if you are running two monitors, you vid card might be a resource hog. We covered a program called "power strip" a while back that lets you de-prioratize your vid card, along with other unimportant resources, and make the CPU pay more attention to what you want it to. Helped me a lot with my FW 410.
-k
Click to expand...

I will post the specs when I get a breather (have to dig out the paperwork). But, I agree, my latest suspect is the display card. I'm considering replacing it, but I'll will try to find that software first! I'd rather not give up our computer long enough to get the video card changed and we're just not PC techs enough to do it ouselves, I"m afraid (we're scared, anyway).

We did manage to move the PCI card for the Delt1010 around until we found a place where it doesn't appear to be sharing an IRQ, but it's still definitely not in the optimum range recommended by M-Audio (IRQ 15 or below).

I am supposed to be programming for someone else right now... ah, addicted!!! Anyway, will try to be responsible adult and continue this conversation sometime next week. I'll post specs.

I will continue this thread because someday it might help someone else, and I very much appreciate everyone's time in attempting to suggest solutions.

Thanks for your help!
 
TexRoadkill said:
No. If there was no audio at that frequency then there is nothing. Nothing does not equal pops and clicks. It is just nothing.

If I were you I would just work in 44.1khz and see if that works okay.
Click to expand...

It does help, a lot (not totally, but it's better enough that we can ignore the random ticks and clicks most of the time).

But we still really need to work this out, since we want to be recording at 96/24, which is why we configured our super duper PC (maybe TOO super duper?) and purchased the software and soundcard that said they could do it. It would be pretty crazy for us to give up (though lately I have been having wild thoughts about trying to sell the whole rig to someone else and get a nice cheap little computer and use Audacity... nah, not yet, but maybe someday.)
 
TimOBrien said:
The basic problem is that you are trying to stream a large amount of information and not all computers are up to it.

Windows machines (no flame I use a P4) are a kludge of mismatched parts from many different manufacturers and not all are in tolerance with each other. Add the inconsistencies of various flavors of Windows and its updates to various manufacturer's drivers and programs installed and you have "little problems" to solve.

I was lucky. My current DAW is a pure Intel mobo with an Intel P4, high quality memory, Win2k SP4 and fast drives with a good stable (Echo Mia) soundcard. My first test was 35 stereo 24-bit/44.1khz tracks- 15% CPU useage and I have ZERO problems.
Click to expand...

I wanted to go Intel, but everyone assured me that AMD was fine. I decided I must be carrying old fashioned prejudice against non-Intel math coprocessors from my ancient past and let it go.

Our computer is supposedly optimized, but I suspect the display maybe too sophisticated and getting in the way. Yes, I wish I had the time and fortitude to break my Windows addiction, it's as bad as smoking tobacco, I suspect. But the effort required is HUGE.

Will post specs next week sometime.

I do watch the CPU usage and most of the time, it's fine, well within tolerance. I have my latency set to the highest latency, safest for recording. But the noises and behaviors during recording do remind me of what happens when I have the latency set too low. The more people who weigh in, the more I"m back to thinking it's the processor not keeping up.

Thanks for the input!
 
Do you have your operating system fully optimized for audio?

Do you have a ton of RAM?

And lastly, if you're using a Delta 1010, what is your DMA Buffer size set to? Try the next highest setting, and see if that helps any. Try the highest one possible and see what that does.
 
A General rule thumb is to set your sample rate at double your highest freq.


So for example...


if 22khz, then a sample rate of 44000

Also consider the limitations of any given equipment



Lee
 
chessrock said:
Do you have your operating system fully optimized for audio?

Do you have a ton of RAM?

And lastly, if you're using a Delta 1010, what is your DMA Buffer size set to? Try the next highest setting, and see if that helps any. Try the highest one possible and see what that does.
Click to expand...

I have taken 90% of recommended optimization steps M-Audio tech sent me to (some I did bypass for considered reasons). My DMA buffer setting on the Delta1010 is at the max available in the drop-down list. We have something like a gig of ram. the machine is dedicated to this, not on the Internet; doesn't even have a printer hooked up.

I am beginning to think I need to explore different direction than processor being overloaded, or too little ram. (See my postings earlier about the cricket experiment. That was definitely NOT overloading the processor or ram or harddrive! But it sure was a very high frequency sound. And the only way it wasn't accompanied by a click on EVERY peak frequency was to drop the recording back to sampling rate of 44.1kHz and bits to 16. It didn't even work to record at 44.1/24 bits.)

I am now going to explore the myriad combinations of settings between the Delta1010 and HomeStudioXL24 to see if maybe I have some conflict there, some setting I don't understand well (the majority that is).
 
Teacher said:
sounds like a delta 1010 or S/W issue,

DId you try format reinstall?
Click to expand...

Could be the Delta 1010. I'm gonna keep working with their tech line. I wish they had better online documentation for that card and monitor/mixer software and control panel.

But, in addition, it is conceivable to me that one of the many settings are screwed up (suggested by forum member treeline, to give due credit).

What do you mean by format reinstall?
 
LRosario said:
A General rule thumb is to set your sample rate at double your highest freq.


So for example...


if 22khz, then a sample rate of 44000

Also consider the limitations of any given equipment



Lee
Click to expand...

Given that, at least in theory, the highest frequency we can hear is about 20kHz, it would seem sufficient then to just go to 44.1. But, I figure there must be some advantage to getting higher frequencies into the recording even if most of us can't necessarily hear it, or the pros wouldn't bother with higher sampling rates.

Here's my problem: if I want to record a cricket, and I try to set the sample rate at 96kHz, and bit rate at 24bits, I get a click on each cheep. If I record at 44.1/16bits, no click, though there is a slight amount of ticking and clicking that I suspect is all that we can still hear of this clicking I'm getting. It's there, but not regular anymore.

When the clicks happen on a high-sample-rate recording, they are so regular as to be impossible to ignore. At least lowering the sample rate seems to make them less frequent and more random,
 
Conversion between digital and analog domains require filters to remove out-of-band artifacts. The farther away from the highest audible frequency you can make the sampling frequency, the simpler (read cheaper) the filtering that is required and the less audible the side effects of the filtering are.

There are people who claim that higher sampling rates provide better sampling period but this is simply not borne out by the science behind digital signal processing that I studied in grad school.

Bit resolution (i.e.going to 24 bit from 16 bit) is WAY more important than sampling rate.

The clicking is due to the larger amount of data being moved...96/24 creates about 3.5 times as much data as 44.1/16, it's that simple. Pops and clicks are nothing more than missing samples in the stream. They are missing because the computer can't keep up.
 
You must log in or register to reply here.
Back
Top