| Author |
 Topic  |
|
Lawrence
Ha`aha`a
USA
1597 Posts |
 Posted - 08/15/2007 : 06:19:49 AM
|
quote:
THE ONE FACTOR that enabled me to play along to another track I'd already created while it played back in real time was disabling the "noise cancellation" option on the AC97 control on my computer.
Aha!... Yes, sounds like the culprit to me as well. In order for the AC97 to do this noise cancellation it has to setup a substantial buffer in RAM or hardware (probably RAM) in order to cancel low frequencies. This in turn, will introduce significant latency.
Glad you were able to solve your main problems.
|
Mahope Kākou...
...El Lorenzo de Ondas Sonoras |
Edited by - Lawrence on 08/15/2007 09:24:07 AM |
 |
|
|
Lawrence
Ha`aha`a
USA
1597 Posts |
Posted - 08/15/2007 : 06:31:12 AM
|
quote:
Does it make a difference if you use shielded cable? EMI is everywhere.
Of course it does Wanda! Most all audio cables you can buy are shielded anyway, except for some headphone extension cables and speaker wire. That being said, some shields are better than others so don't expect cheap cables from Radio Shack to have good shielding for instance.
Also common are balanced and shielded cables. All of the "Cannon" three pin microphone cables are balanced and shielded. Balanced cables allow the interfering signals that make it through the shield and are coupled into the two internal wires to be canceled by using cancellation circuit at the preamp end (opposite end as the mic). This cancellation circuit is either a floating transformer (old technology) or the plus and minus inputs of a differential amplifier. A differential amplifier is an amplifier that only amplifies the difference between two signals, and the transformer works the same way. Since the interfering signal is the same on the two balanced wires, it will be ignored by the transformer or differential amplifier.
Professional audio equipment also uses balanced cables for almost all signals, not just the microphone ones.
|
Mahope Kākou...
...El Lorenzo de Ondas Sonoras |
|
|
|
Absolute
Lokahi
275 Posts |
Posted - 08/15/2007 : 07:02:20 AM
|
If you've got the means for professional level equipment, you'll get better results. Most people just want to play around a bit. I attended a lecture by a notable figure sponsored by a professional organization in which it was even pointed out that there is a metallic material with adhesive on one side that can be applied in a "cigarette wrap" around unshielded wires, but that was in relevance to laboratory testing equipment use for radio frequency testing and experimentation. (Note that shielded cables help with radio frequency interference imparted by traveling waves. The wave encounters the shield, and is essentially induced to spread out in the shield. If it hits the air boundary on the inner conductor side, it is mostly reflected back into the shield due to the impedance mismatch between the metal of the shield and the air/insulator material inside the shield. If a current is induced in the shield inductively rather than by a traveling wave, it produces current in the shield. Current in the shield produces flux as well, which then cuts the central wire, to some extent, to produce current and a Voltage along the length of the wire, in an attempt to produce a current to cancel that inducing the incident flux from the shield wire.)
In my experience, pops and crackles materialize only when I permit my microphone wire to get tangled among the others. I try to keep it away from power cords and the CRT monitor, and away from wall outlets (and the wires running to them from behind the drywall that power other loads). Basically I've run the microphone wire at 70 degrees upward from the plane of the floor slanting slightly away from my computer, and taped it to the wall to keep it from becoming entangled again.
E-M fields decrease with the square of the distance, and flux will induce Voltage along the length of the microphone wire if it is lying in the same plane with other wires carrying current (as in a bunch of wires tangled on the floor or behing a computer). This 70 degrees up and away from the computer approach (my computer is under my monitor in a desk arrangment) keeps the microphone wire out of the plane of the other wires tangled on the floor, reduces the amount of its length in proximity to them and the region of strongest E-M fields due to those wires, and moves it away from the point where wires exit the computer and away from the monitor, both noise sources. This is a compromise to keep the microphone away from the monitor noise as well and out of the plane of the other wires.
EMI is everywhere, but if the noise signal level is low enough compared to that of the microphone, it ceases to cause the pops and bursts of crackles. IF the microphone wire is perpendicular to the plane containing the tangle of wires on the floor, flux originating from current in those other wires cuts the plane of the unshielded microphone wire and tries to induce current that flows across the diameter of the wire, rather than along its length. That reduces the magnitude of the Voltage that can develop along its length due to incremental elements of flux originating in other wires, and thus the magnitude of the noise signal that can be imposed by currents flowing in other wires.
|
Thank you. |
Edited by - Absolute on 08/16/2007 03:51:11 AM |
|
|
|
Absolute
Lokahi
275 Posts |
Posted - 08/15/2007 : 1:40:53 PM
|
Audacity is PC based. Shielded microphone cables typically come with a jack that won't fit in a sound card's 1/8" jack.
Shielding helps by reducing the ability of E-M waves to reach the wires carrying the signal that you wish to remain pristine. (The E-M interference wave transmitted through the air is substantially reflected back into the metal of the shield when it reaches the boundary with the internal insulator/air, so very little reaches the wires inside the shield.)
With a pair of wires inside a plastic or rubber insulator, as with my PC microphone, there is no shielding, so they are subject to both E-M wave based interference and inductive linking with other conductors, including power conductors. This can make them prime candidates for picking up electrical noise corresponding to crackle and pops. You should keep them away from sources of such noise, like monitors, power cables, and equipment that generates electrical "noise".
Most people using Audacity are using PC's, and just want to have a bit of fun. If you're using an inexpensive PC microphone without a shielded cord, keep the cord away from noise sources and orient its length as much as possible perpendicular to the floor, where other cables lie strewn. This helps to reduce noise problems. If your cable isn't shielded, untangle it from the other wires strewn behind your PC, and position it so its length isn't in the same plane as those wires on the floor. I taped mine to the wall about three feet from where it leaves the computer with the wire oriented at about a 70 degree angle (almost perpendicular to the floor but slanting away from my monitor, which is above my PC). When I use it, I put it about three feet from the point where its taped up on the wall pointing away from the computer (with its fan noise). This keeps the cord from tangling with others.
With professional grade microphones using shielded conductors (within a cable), a metal shield is formed around the conductors carrying the current, but current can still be inductively induced in the shield conductor by noise sources. That in turn can induce current/Voltage in conductors inside the shield. Fortunately, those internal conductors are twisted pairs. With the twisted pair conductors inside the shield across the terminals of some amplifier input, nearly identical noise signals induced on both internal conductors (of a twisted pair) provides for cancellation of the noise at the terminals of the amplifier (due to differential amplifier configuration, essentially an amplifier with a floating input that then uses one microphone wire as its reference. Only the difference between the input signals is amplified, so a noise signal that is identical at any point in time on both inputs is subtracted from itself when the input difference signal is produced).
You can learn more about professional microphones and shielding here:
http://www.lightsounds.com/index.php/page/page31
(Remember to keep the microphone as far from the computer as you can, within reason, to avoid picking up audible fan noise. A directional microphone pointed away from the computer can help in this regard.) |
Thank you. |
Edited by - Absolute on 08/16/2007 04:00:15 AM |
|
|
|
Absolute
Lokahi
275 Posts |
Posted - 08/16/2007 : 04:14:56 AM
|
| My AC97 audio chipset driver noise cancellation option is software that Audacity doesn't know about. That noise cancellation feature has to induce a time delay to work. That's not a by-product of buffering built into the OS and system architecture and the time required for data to pass through multiple layers of buffers. The delay I described results from the artificial introduction of intermediate software between the point at which an audio signal is captured and the point at which Audacity is supplied with that signal that inherently introduces a delay due to its prescribed mode of operation. You can't capture a signal then subtract a noise signal perceived as underlying an audio stream based on past data inherent in that audio stream in real time without inducing a delay as you monitor for then subtract noise as it occurs at different levels. Audacity simply doesn't know that the delay is being introduced by the AC97 software driver that is trying to cancel noise. I can't control delays due to buffering processes. They're part of the OS and system architecture. I can control whether I introduce a time delay due to use of a noise cancelling program between the microphone and Audacity's receipt of the signal from an intermediate processing routine. Those encountering problems with multi-tracking might wish to check to see if they too have introduced some intermediate processing through a sound card driver feature that is causing this delay, such as this "noise cancellation" function I had selected, and experiment to see if unchecking that option eliminates a time delay problem with multi-tracking under Audacity, rather than assuming that the problem is due to a latency issue due to multi-layer buffering processes inherent in the OS and system architecture that they can not control. |
Thank you. |
Edited by - Absolute on 08/16/2007 04:17:59 AM |
|
|
|
Lawrence
Ha`aha`a
USA
1597 Posts |
Posted - 08/16/2007 : 07:02:12 AM
|
quote:
Those encountering problems with multi-tracking might wish to check to see if they too have introduced some intermediate processing through a sound card driver feature that is causing this delay, such as this "noise cancellation" function I had selected, and experiment to see if unchecking that option eliminates a time delay problem with multi-tracking under Audacity, rather than assuming that the problem is due to a latency issue due to multi-layer buffering processes inherent in the OS and system architecture that they can not control.
Actually most "professional" audio software and also the drivers for professional sound cards DO give the user control over buffer sizes. (just check the support forums for Adobe Audition, or Reaper, or Protools, or Sonar, etc., etc.).
In addition the noise cancellation that your AC97 implements requires the use of a (additional) buffer to store an average value of the repetitive noise so that it can be canceled. You cannot perform wide band noise cancellation (multiple non-harmonic frequencies) with just a delay, the intermediate samples have to be stored (in a buffer of course). That is why turning off the AC97 noise cancellation sped up your process and reduced the latency. Certainly, experimentation on these kinds of issues are important, but it is a failure on the part of Audacity to have not provided direct control over some of the buffer sizes as does most "pro" software.
No matter what kind of Audio software and hardware you use, unless you buy a pre-packaged and "tuned" system, you will have to tweak many things.
Also available at the Adobe forums and Reaper forums (and possibly the Audacity forum) are list s of things you can do to Windows to improve your audio performance, including such things as turning off useless animated effects that can steal up to 99% of your processor time!
TweakUI is one free tool you will definitely want to download from Microsoft that enables improved audio performance.
|
Mahope Kākou...
...El Lorenzo de Ondas Sonoras |
Edited by - Lawrence on 08/16/2007 07:36:41 AM |
|
|
|
Mark
Ha`aha`a
USA
1628 Posts |
Posted - 08/16/2007 : 07:07:57 AM
|
Hi all -
Latency is a fact of life with computer based recording. There are many ways to get around it; as Absolute mentioned, disabling software processing at the input will help. So will using real software instead of free-ware, disabling FX, adjusting the buffering, cutting down the number of playback tracks, buying a faster computer, using a low latency interface, carrying a lucky rabit's foot, adjusting the flow on the lava lamp, etc etc etc.
One EZ tip: when I'm overdubbing, I simply mute the playback on the input track and listen with one ear to what I'm playing and one ear to the 'phones (old studio trick...)
Here's some more info on latency and what to do about it: http://www.sweetwater.com/insync/techtip.php?find=01/06/2005
BTW: I'm afraid I can't see any situation where you'd want to de-noise your tracks at the input stage. If your recording environment is that dirty, clean it up first.
Otherwise, I'd suggest saving noise correction (and any other processing) until you've got all the tracks recorded. Why? For one thing, whatever small amount of noise is on the track may be masked by the other tracks. And, secondly, consumer-based noise reduction is almost always problematic and may actually make your tracks sound worse if applied with a heavy hand.
Happy recording. |
|
|
|
Lawrence
Ha`aha`a
USA
1597 Posts |
Posted - 08/16/2007 : 07:13:21 AM
|
quote:
BTW: I'm afraid I can't see any situation where you'd want to de-noise your tracks at the input stage. If your recording environment is that dirty, clean it up first.
Otherwise, I'd suggest saving noise correction (and any other processing) until you've got all the tracks recorded. Why? For one thing, whatever small amount of noise is on the track may be masked by the other tracks. And, secondly, consumer-based noise reduction is almost always problematic and may actually make your tracks sound worse if applied with a heavy hand.
Excellent point Mark, denoising AFTER recording is almost always best, provided you have taken all the usually appropriate steps (shielded cables and such) with the gear.
Some of these built-in sound chips (like the AC97 Absoulute mentions) are placed on a relatively noisy main computer circuit board, hence the unwanted induced noise. Kind of a neat trick by the designers of the AC97 to try to reduce this local interference, but of course it causes latency. Better to use FFT-based repetitive noise reduction after the fact even in this case.
|
Mahope Kākou...
...El Lorenzo de Ondas Sonoras |
Edited by - Lawrence on 08/16/2007 07:31:32 AM |
|
|
|
Absolute
Lokahi
275 Posts |
Posted - 08/16/2007 : 3:00:03 PM
|
Definition of "latency" may be found at:
http://searchsmb.techtarget.com/sDefinition/0,,sid44_gci212456,00.html.
My perspective in prior posts is based on latency as a factor produced by an OS and motherboard architecture and buffers controlled by the OS and CPU of a computing system as part of normal overhead, not my intentional introduction of an intermediate program that must store data because it can only function by looking backward in time at what it regards as noise in the recent past to attempt to approximate the most recent level of noise that must be eliminated.
The definition used by others here seems to focus on any delay for any reason in producing audio data that can be used by a program, which seems consistent with more general usage of the term.
The manufacturer of Audacity (or presumably any other PC recording suite) has no way to know that I'm going to use an external program paired with the audio chipset that is going to have to force a time delay because it needs to sample half a second's worth of data before it can start feeding the noise reduced result to some other process that requires it.
My reference to "professional" equipment producing better results was general, and was actually in response to an inquiry regarding shielded cables. I make no assertion regarding professional audio cards. I am using Audacity, which was the program with regard to which an inquiry was originally made at the start of this thread. That's why I responded. I began by noting that I am responding with what I discovered with my very basic system for others like me doing recording for a bit of fun on a basic home system. (I now have no problem using this basic home system for single tracks as long as I exercise a few precautions regarding the microphone cord. I just identified the sources causing the problems with multi-tracking on my system, and now the two problems I was facing, one likely due to a recording speed mismatch between Audacity and Windows' control panel settings, and the other due to an audio driver induced delay, is resolved as well, both through basic experimentation.)
I have NO complaint regarding Audacity (except that its doesn't do perfect noise clean up, and fix all my non-ideal environmental problems without any input from me whatsoever, so I can have perfect, effortless CD quality recordings in a noisy household environment, which is clearly nothing to complain about.) The creator of Audacity noted that he is trying to write it for everyday users. Providing options for buffer size manipulation could intimidate a lot of average, home users. (He also gives his fine product away, so he deserves to be defended.)
I get virtually no noise (that I can associate with electrical sources, such as pops and crackle) in a recording if I'm careful with my microphone cord's placement relative to other cords, the monitor, and other electrical noise sources, even with an audio chipset on the motherboard. That one factor is the difference between recordings I can live with and those I'm inclined to delete, at least with regard to identifiable electrical noise.
With the microphone cord untangled and oriented almost vertically, slanting away from my computer over the first three feet (at which point I tape it to the wall), and my incoming audio processing software disabled to eliminate time delays when multi-tracking, the only possible improvements I might consider would be a quieter computer fan. I think Audacity is great stuff. I've got Music Creator as well (from the Cakewalk people). When I try to record using it, I get MUCH more noise on the recording than with Audacity. (I don't know why. If Audacity is working, Music Creator should do at least as well, since it is one of those bought and paid for software programs that is inherently superior to freeware.) |
Thank you. |
Edited by - Absolute on 08/16/2007 3:14:02 PM |
|
|
|
Lawrence
Ha`aha`a
USA
1597 Posts |
Posted - 08/17/2007 : 07:14:01 AM
|
quote:
The creator of Audacity noted that he is trying to write it for everyday users. Providing options for buffer size manipulation could intimidate a lot of average, home users. (He also gives his fine product away, so he deserves to be defended.)
Actually, They are not a He, they are a group of developers organized under the SourceForge Open Systems Foundation. There are many hundreds of projects being developed thanks to the OSF. Most of the SourceForge Applications (including Audacity) are designed for a Linux environment and are essentially "ported over" to work under Windows. I considered joining the development team myself to develop some "plug ins" for Dolby and dbx decoding for Audacity. I use several of the OSF / SourceForge tools in my day job.
For what it is, yes it is a fine product, just needs a bit more development to place it in the same group as Audition or Sonar or Reaper. I agree, just because software is free, does not automatically mean it is worse than "paid for" stuff. There is a lot of good quality software that is also freeware.
Here are the more involved current Audacity team members (I did not copy the whole list).
Audacity Team Members
* Gale Andrews, documentation and support
* Richard Ash, documentation and support
* Christian Brochec, documentation and support, French
* Arturo "Buanzo" Busleiman, system administration
* James Crook, developer
* Vaughan Johnson, developer
* Leland Lucius, developer
* Dominic Mazzoni, developer
* Markus Meyer, developer
* Alexandre Prokoudine, documentation and support
* Martyn Shaw, developer
Technical leadership council
* Dominic Mazzoni - (dominic (at) audacityteam.org)
* Matt Brubeck - (mbrubeck (at) audacityteam.org)
* James Crook - (james (at) audacityteam.org)
* Vaughan Johnson - (vaughan (at) audacityteam.org)
* Leland Lucius - (leland (at) audacityteam.org)
* Markus Meyer - (markus (at) audacityteam.org)
Emeritus: distinguished Audacity team members, not currently active
* Matt Brubeck, developer
* Roger Dannenberg, developer
* Joshua Haberman, developer
* Monty Montgomery, developer
* Shane Mueller, developer
* Tony Oetzmann, documentation and support
Contributors
* Lynn Allan, developer
* William Bland, developer
* Brian Gunlogson, developer
* Arun Kishore, developer
* Harvey Lubin, graphic artist
* Greg Mekkes, developer
* Abe Milde, developer
* Paul Nasca, developer
* Augustus Saunders, developer
* Mike Underwood, developer
* Jun Wan, developer
* Tom Woodhams, developer
* Wing Yu, developer
|
Mahope Kākou...
...El Lorenzo de Ondas Sonoras |
Edited by - Lawrence on 08/17/2007 07:18:58 AM |
|
|
|
Absolute
Lokahi
275 Posts |
Posted - 08/17/2007 : 11:46:38 AM
|
| Here here! |
Thank you. |
|
|
|
Lawrence
Ha`aha`a
USA
1597 Posts |
Posted - 08/17/2007 : 2:01:41 PM
|
One more note on this subject (pun intended).
Out of curiosity I downloaded the most recent version of Audacity (1.3.3-beta). In Edit->Preferences->Audio I/O they have added two Latency setup parameters. "Audio to Buffer" and "Latency Correction". The default for Audio to Buffer is 100mS, but probably can be reduced, and the default for Latency Correction is Zero mS, which is not realistic. The procedure I mentioned far above can be used to measure the Latency Correction value after you have settled on a suitable Audio to Buffer value. Usually the Latency Correction is pretty close to the same as the Audio to Buffer but sometimes can be longer.
So now the Audacity folks have joined the rest of the audio recording community in regard to Latency Setup.
|
Mahope Kākou...
...El Lorenzo de Ondas Sonoras |
Edited by - Lawrence on 08/17/2007 2:03:42 PM |
|
|
|
Absolute
Lokahi
275 Posts |
Posted - 08/20/2007 : 08:19:50 AM
|
And if you would like to get either the current or the beta version of Audacity, you can find the download page here:
http://audacity.sourceforge.net/ |
Thank you. |
|
|
|
hwnmusiclives
`Olu`olu
USA
580 Posts |
Posted - 08/22/2007 : 06:53:43 AM
|
To all who attempted to assist me with my latency problems, mahalo. I thought I would give you an update. I haven't quoted any of your solutions here because they were lengthy - and none of them applied to my situation. ;-)
As it turns out, when I installed a new soundcard, I did not disable the motherboard's native soundcard. (On some PCs this needs to be done through the BIOS; on others, by flipping a switch on the motherboard.) So while I had a new soundcard plugged in, the PC was effectively using the original soundcard for input. But I configured my recording software to pick up signal from the new soundcard. So, essentially, by not disabling the original soundcard, I was using both simultaneously and they were competing desperately for the same system resources. It is also entirely possible that the second soundcard configured for my multitrack recording software was picking up its signal - late! - from the first soundcard. How's that for messing up the signal path?
Thanks for all the great advice. I'll add to it my own newfound wisdom: When you upgrade your soundcard, be sure to disable the old one.
|
Join me for the history of Hawaiian music and its musicians at Ho`olohe Hou at www.hoolohehou.org. |
|
|
|
Lawrence
Ha`aha`a
USA
1597 Posts |
Posted - 08/22/2007 : 07:51:53 AM
|
quote:
As it turns out, when I installed a new soundcard, I did not disable the motherboard's native soundcard.
This is not always necessary, sometimes they can co-exist depending upon how well written both the old and new sound drivers are.
P.S. : Soundblaster products are some of the worst at being incompatible, stay away from them if you can. they also mess with sample rates (without telling you) and fail to say in sync with many other things, including themselves. They are OK for game soundeffects but not much else. If you already have one, my condolences. I have one as well, but I won't let it anywhere near my DAW (Digital Audio Workstation) computer!
At least most everyone reading this thread knows a little more about Latency! (or else they fell asleep).
I noticed the original poster (Mark E) has not responded in a while.
He must have solved his problem on his own as well.
|
Mahope Kākou...
...El Lorenzo de Ondas Sonoras |
Edited by - Lawrence on 08/22/2007 07:54:26 AM |
|
|
|
Topic |
|
|
|
Topic Locked
