I have some “light” clicks that are not easily spotted in the waveform (have to zoom in real close). The “Click Removal” effect is not able to get rid of them. The “Repair” effect can (they are very short), but it is very time consuming to identify the “broken” waveform and would not be practical to address several hours of audio.
So the question is why is this occurring in the first place? I have a laptop connected to the Scarlett Solo, which is connected to another preamp. My guess is that it is because it is a cheap laptop, but that is just a hunch.
Are you getting [u]dropouts[/u]? That’s usually the only thing that goes-wrong “digitally”. Otherwise, “noise” is an analog problem.
My guess is that it is because it is a cheap laptop,
Usually dropouts are related to multitasking, and your operating system is always multitasking, even if you’re only running one application. A faster computer can often help but that’s NOT the root cause. [u]Troubleshooting dropouts[/u]
I don’t think it is a dropout. Here is a snapshot of the waveform before and after the repair. You can see something gets weird in the before, which translates to a click. But you have zoom in real close to even see it…it is just 0.0009 seconds long and there is not associated spike, so it can’t be easily identified.
Do you get these spikes with silence? If yes, I’m blaming something analog. Is it related to recording level at all?
You see this immediately after recording, right? Not after editing or mixing or applying effects?
…It almost looks like an “overflow” with “bad programming” but you’re NOT clipping so I don’t think so. …With signed binary integers the most-significant bit is used for the +/- sign. If you “count” too high (as with clipping), the number-bits get carried into the sign bit and the sign-bit can be “flipped”. And because negative values are represented in something called two’s compliment, it’s not just a change in the +/- sign, the data gets totally fouled-up.
DSP programmers are aware of this so you rarely see it and I’ve never seen it with Audacity. In fact, I’ve never seen it with any audio program but I have seen it with another application and I didn’t know what was happening until the programmer found the bug and explained it.
Here is a zoomed out view of the waveform (about a half second). I highlighted the general range and you can’t see anything off at this level. The affected waveform is only about 1000th of a second and its not really a “spike”. You have to really zoom in to even see something is off.
Regarding the silence, let me take a closer look at the raw audio. I am looking at the post processed (noise reduction, de-click, equilization, noise gate, de-esser, etc)…heck, maybe something happened during the processing. EDIT: I had a listen to the raw file and the clicking exists there as well. So it is not something that is added in during processing.
I did find a workaround though. The audacity manual points to a “ClickRepair” program, which can address this in bulk. So far that seems to work. It would be nice to find the root cause though so I will look at the raw file and see if its in that too.
