Are there techniques to heal dropouts?

Recently I have started getting very bad problems of latency. These may be connected to the Windows update that includes the patches for Meltdown and Spectre (and possibly specifically related to the fact that I have an AMD processor), because I can't think of anything else that has changed on my system. This means that I now have to do many more takes than before to be reasonably sure of getting a usable recording on a track. But sometimes I find that I have to throw away a take that I like and that would be hard to repeat in exactly the same way, just because of one small pop. Are there techniques that might sometimes be able to rescue a track with a dropout?
 
If you can't copy/paste or time-stretch and x-fade over it, Izotope's RX (full product) can do some pretty amazing things. It's pretty expensive though, so I'd go through some optimizations, and make sure you're disabling unnecessary/draining plugins (maybe you have a low-latency mode for recording in your DAW), and even print/freeze anything that you know is taxing the system.
 
Thanks very much for replying.

Izotope RX is a bit outside my budget right now, but if I have correctly understand the principle of the technique you are alluding to, the kind of thing I would be looking to do to do the repair on the cheap would be to slice at the point of the dropout, pull the earlier part of the item a bit to the left, time stretch both ends a little bit, and then push the earlier part back to start where it started before, so that the two new items now overlap. Have I got that right?
 
..., but if I have correctly understand the principle of the technique you are alluding to, the kind of thing I would be looking to do to do the repair on the cheap would be to slice at the point of the dropout, pull the earlier part of the item a bit to the left, time stretch both ends a little bit, and then push the earlier part back to start where it started before, so that the two new items now overlap. Have I got that right?
Short answer, "yes."

I don't know which DAW you're using but in Logic Pro X I can set the point at which the time stretch starts within a clip, so when I "pull" it doesn't move the entire clip, but just as much of it as I want, i.e., no realignment at the other end is required.

It really depends on the size of the dropout, and what the track's prominence in the mix is at the point of repair, but it's worth a shot, assuming you can't just punch in that little piece and x-fade that in, which is the other tactic that I'd try before re-doing the whole thing. (Assuming you've got the resource/buffer/whatever issue fixed to insure a new take wouldn't have the same problem!)
 
I think we're on the same page. I just wasn't expressing myself very well! I'm using Reaper, so maybe I should ask in the Reaper forum for the most efficient way of doing this, but I think I've got the general principle now. I've used time stretch to fix bad timing in my bass lines, so I know how that works. Cross-fade was the word that I couldn't remember in my previous post!
 
Gads, chase the issue of the non stable recording!

Yea, this^^

It could be on a magnetic hardrive that it is getting full and it is storing the data on the outside of the disk causing it to slow down. Probably more noticeable if the drive is 5400 RPM.

But a drop out and a pop are not the same thing if I understood correctly.
 
Pop in recording-crossfade where noise is. Drop out, like a tape drop out, would show up as a flat line where there should be a waveform. Whole different issue, and usually cable problem somewhere.
 
Recently I have started getting very bad problems of latency. These may be connected to the Windows update that includes the patches for Meltdown and Spectre (and possibly specifically related to the fact that I have an AMD processor), because I can't think of anything else that has changed on my system.

Based on tests some people have done...the patches have showed no effect on DAW processing power.

You need to do some detective work and find the cause. I know often people say "I have changed anything"...without realizing that they may have made a bunch of changes to their overall process of how they work...so it may not be just a system change...but even that sometimes happens almost without notice, some configuration adjustment, etc...that's not noticed immediately.

The best way to handle a dropout...tell them to get a job and start paying rent, or get out.
 
I've been tearing my hair out trying to find out what is causing the latency. In the meantime I need to find a way to keep working at my stuff.

The disk isn't getting full, so that won't be slowing things down. Latencymon gives me various culprits such as ntoskrnl.sys, ataport.sys, usbport.sys (even when there is nothing plugged into USB) and various things to do with graphics. That doesn't help me very much.

I can't see how my working methods have changed because they haven't. I'm using the same software as before.

When I said that I suspect the Spectre patch, I wasn't talking about a general slowdown because I know that AMD processors aren't badly affected by that. I was wondering whether the patch had caused something else to happen. I've already read that the patch for Windows has made some older AMD-based machines unbootable, so what I suspect isn't necessarily absurd.
 
Not saying it was absurd...just that on a couple of DAW app sites...people have install the patches, run tests, and not found it was impacting processing.
Could patching have done something else to you computer...yup...so you can open Control Panel and go to you installed programs...and then check the install updates link there, and sort by date. You will see what MS has put on your computer recently...uninstall it and see if it fixes things.
You can always run MS update and let it put patches back.

On my two online computers...I have updates/patches set to only notify me, so I can monitor what's being installed...and I have not yet gotten those specific patches from MS for the Meltdown and Specter bugs. Not sure how they are rolling them out and in what order....maybe based on most current OS first...?...I'm running Win 7 on all my systems still...and maybe the different OS's respond differently to the patches.

Thing is...you're talking latency...which for basic recording, wouldn't be really a CPU intensive issue, unless you are trying to also run a bunch of real-time FX/processing on the tracks while recording...?
Most of the basic latency issues tend to be an interface and ASIO buffer related kind of thing...or the hard drive just can't keep up, and you get dropouts...but it would have to be a real dog drive, or you trying to record a few dozen tracks at very high sampling rates.

Can you provide more specific details of your system...buffer settings...interface settings...etc...and what/how exactly are you recording?
Not to be a nag...but really trace out all your steps and settings and see if you can spot something that you might have changed or that is different this time around.
 
I'm pretty sure the problem is not to do with my audio interface or ASIO specifically, nor how many tracks I am using because Latencymon still reports very high peaks of latency even when the DAW is not running, the audio interface is unplugged and I am not sitting at the machine doing anything with it. At these times the only hardware I have connected is the processor, the motherboard with ethernet and on-board sound (which I never use), the graphics card, and a USB wireless keyboard/mouse receiver.
 
Pardon me if I am being a PC numpty (often is!) but I thought the latest security scare and patches only related to Intel processors?

Otherwise I agree, get dropouts sorted. (I bet Samplitude Prox3 and Soundforge between them have something to do it? I know SF has a pretty good 'clipping fixer')

Dave.
 
Google "ntoskrnl.sys ataport.sys usbport.sys latency" and see if any problem solves appear.

The trouble with this is that they are only very vague indicators of where the problem might lie. ntoskrnl.sys is the main Windows kernel, so that could mean anything. I can find all kinds of things to try in various forums, but few of them seem to be suggested with great confidence. And when the drivers implicated are related to every aspect of my computer from graphics to SATA to USB to ethernet and even to Latencymon's own sys file, that means that I have a lot of different settings to tweak randomly and hope that I don't make things worse. When my bike isn't running well I don't just start turning bolts willy-nilly. I need to have a clear indicator of where the problem lies before I risk doing anything. The same goes for my computer. I don't want to end up having to do a clean install of Windows because I messed up my machine by blindly faffing about with it.
 
Well, this is really the subject of a different thread, but the clicks and pops are not latency per se, but a resource problem, usually meaning a block of data got dropped. So, something has changed that means the system is not keeping up. Possible things to test to narrow down causes:
  • Do a single track of a single instrument, like a guitar, with no FX on a low sample rate and bit depth and see if that has the problem.
  • If you have a problem at the lowest case, then you probably do need to consider a rebuild, at least do a full backup because that kind of quick degradation can indicate a lurking hardware glitch. I'd clean start, or maybe run some disk/system benchmarks to make sure you don't have a hardware problem. And that's a good time to move to newest tech SSD if you haven't already.
  • If no problems with the base case, then do it again increasing bit depth, sample rate, tracks, until the problem surfaces. Try fiddling with buffer size, or moving the project to a fast external drive (USB 3.0 or faster) to see if that makes a difference.
  • Try the same thing but adding FX to the tracks, cycling through the common ones you use and see if one of those is a problem.
 
(Oops, meant to edit, but quoted myself :facepalm:... Too early... Quote deleted)

P.S. I assume you are doing all of this while disconnected from the network, anti-malware disabled, i.e., the normal optimizations done.
 
Back
Top