The discussion got all the way down to screws, nuts, and bolts but then petered out before an actual answered was developed.
Somehow, if you apply Noise Reduction in a Macro for a fresh show, Noise Reduction will generate a useful Profile. I also believe if you apply Noise Reduction manually and just skip the profile step, it will do the same thing.
I agree completely that producing a Profile ahead of a Macro or series of Macros is desirable and will give Audacity something to chew on, but apparently, that’s not absolutely needed.
There were some suggestions posted about sensing quiet portions of the show for Profile generation, or going for the first and last segments of the show, etc. I don’t think any of that is the answer.
I think someone was alone in the lab one dark night and discovered that if you averaged out all the tones in a story, the result was a not perfect, but serviceable Noise Reduction Profile.
That would give you a profile of “the story”, which is probably the part that you want to keep.
You could probably make some generic noise samples for different use cases, such as tape hiss.
Maybe not. Noise is there 100% of the time. My voice is there some lesser percentage. That right there might be useful. Also you could intentionally gate out sounds over a certain volume and you could tailor the gate depending on the show. I’m going for the goal of no human intervention because that’s how it works. Select the show, apply noise reduction, go home.
Dominic’s original version of “Noise Removal” (as it was then called), looked for low points within the selection, then looked for spectral peaks in those low points. (I recall a code comment that joked about “the max of mins”). The idea was to implement an efficient way to estimate the noise floor spectrum without having to perform a full analysis of the entire track (which could take a long time).
I think Paul’s version (now called “Noise Reduction”) takes a simpler approach and assumes that the selection for the “Profile” step is representative of the noise floor.
I’m not clear whether you are asking about how to use the Noise eduction effect that we have, or how a better effect might work.
No. It won’t. At least in 3.4.2, you never get the OK button. It’s gray.
Which brings us back around to: What if there is no profile step? You just turned Audacity on, the clip is fresh, and you apply a Macro with Noise Reduction in it.
I’m beginning to generate a Conspiracy Scenario. If you apply Macro Noise Reduction to a fresh clip, it does nothing, but it doesn’t fail. The user goes away with a fuzzy-warm feeling and everybody wins. Would you know if your background noise was 6dB quieter without ripping it apart and analysis?
There is a sister forum post who wants to pile his fixed laundry list of effects into a Macro. Noise Reduction was the holdout.
If the Macro is applied to the current project, the current selection is used to create the Noise Profile. If the Macro is applied to files, the first file (all of it) is used to create the Noise Profile.
That sure sound a lot like, “Select everything, cross your fingers, and hope for the best.”
My goal is to get a feel for shows which may fail or do something unexpected.
If I recall correctly, it does say in the manual that Noise Reduction is not well suited for macros.
Also, the Noise Reduction effect is a relic from the past. It is the only remaining effect that inherits from the old effect base class. If you recall Leland updated all of the built-in effects except for Noise Reduction (which Paul was working on at the time). It has been ripe for replacement for a long time, so there is zero incentive for a developer to spend time making improvements to the current effect.
I’m sure that it will be replaced eventually, but in the meantime my recommendation is to do noise reduction manually (not in a macro).
I can make this worse. I had a recent short sound presentation that, when I applied audiobook mastering, appeared to be quieter— have less background noise.
I did not apply Noise Reduction, but had I had Noise Reduction in the Macro, I’d be willing to bet a lot of chocolate that it worked. See? It got quieter!!