Help identifying the cause of the "pop"

Still hoping Trebor chimes back in on the Spectogram.

Did you flash the bat symbol?

If I didn’t seem as excited as I meant to be, I revised the graphic. See red dots. Your microphone is doing this pop thing all the time, it’s just some of them are a lot louder than others.
Crackle-Pops.png
Koz

So I see! Yikes! I did try a recording on my desktop machine. The pop was there, as you would have expected based on the above, but much less noticeable except for one time.

Bat symbol?? I think this one is as close as I can come: :smiling_imp: I might try a message, but I hate doing that because everyone is so gracious with their time, I don’t want to impose. Then again, with so many messages, one or two can get lost along the way.

Still, if I could fix 3-4 really noticeable ones, it could tide me over until I get a new microphone. :slight_smile:

I might try a message, but I hate doing that because everyone is so gracious with their time

Go ahead and post a message. Tell him I sent you. We need to get you all the way to a finished production and there’s still pieces missing.

Koz

Thank you. I’ll do it!

There is a DeClicker plugin for Audacity here , it can automatically remove pop/click noises.
However that DeClicker plugin is slow : it takes longer than playback time to process the audio, but it’s faster than removing them manually.

DeClicker settings to remove cork-pop noise.png
DeClicker settings to remove mouth clicks.png
Even after both those DeClicker settings have been applied, your audio still needs noise-reduction & DeEssing , IMO.

Traveling, but will read again, try (as soon as I get a good connection), and report back.
Thanks so much!
JB

Trebor–
I’ve given this a try. Some clarifications, please.

  1. Should I apply Declicker to the whole file, or only where I hear a pop? I’m thinking the whole file, because there are probably some subtle ones I don’t hear.

  2. I used the settings in the first screen shot on a selected, small area where I knew there was a pop. The pop was fainter, but there.

  3. I used the settings in the second screen shot to do it again on the same area. I still think I can hear it, but barely.

Was this correct? Use screenshot #1 on the first pass, and if not totally successful, use shot #2 to try again? That is, I’m not sure when to apply each screen shot. Perhaps I wasn’t as successful because I didn’t select the whole file?

  1. In theory, should these settings work on all of my pops? I didn’t try it on the file I posted that you corrected, but I’ll do that next. I’ll post my results.

  2. I thought I had applied noise reduction to the file I posted, but maybe not, or maybe I didn’t do it correctly. I selected a silent area, chose Noise Reduction and clicked the Get Profile button. I couldn’t do anything else at that point without selecting Noise Reduction again, so I immediately selected the whole file, then Noise Reduction, then used the defaults there and clicked OK. Is that right?

Thanks for your time and patience!
JB

Either way will work. If you’re applying it manually, select a region of ~1/2 a second, with the pop in the middle.

Try increasing the “Number of Passes” setting from 1 to 2 , (this will increase processing-time).

I suggest always making two passes: apply each of the two different settings shown once.
The different settings reduce different types of noise.
You can save the settings as two presets in the plugin to save you from having to input them repeatedly …
Can create presets in the DeClicker plugin.gif

No guarantees the plugin DeClicker will catch everything : it’s not as good as doing it manually, but it is much quicker.

That sounds right. Applying noise-reduction is a 2-step process : Step #1 capture the noise-profile of a region which should be silent, (once captured, Audacity will remember it until Audacity is closed). Step #2 select the region where you want to reduce noise and apply noise-reduction like any other effect. I suggest noise-reduction values 6 6 6 …
Subtle noise reduction settings 6-6-6.png

Thanks for this added info, which I’ll go back and digest.
I think I did the Declicker wrong… I figured out my mistake and the second go is a winner, or at least a runner up. At the time of this posting, I only made one pass with the first set of settings.

I’m posting the original sample with the pops, which I normalized
Then the file after running Declicker
Then the file after running Declicker and DeEsser. I can’t hear much of a difference with the DeEsser. I used default settings.

JB


The two different settings reduce two different types of noises :
this one, (first set), removes pops , this one, (second set), removes clicks.
I would use both on your audio. If I was limited to one I’d use the pop settings.

BTW You should make your stereo tracks mono* by discarding one of them because …
#1. It halves the processing-time.
#2. DeClicker & DeEsser plugins can create a distracting pseudo-stereo effect on stereo tracks.
#3. The audio file-size will be reduced, (less computer memory & quicker to transmit).

[ * Choose “split stereo to mono” then delete one of the tracks by clicking on the “X” on the far left top corner of the track].

Got it.

I’ll be working on my files over the next couple of days. You suggested I make the tracks mono. Since I’ve already done noise reduction and normalization on half of them, to stay ‘parallel’ should I continue in that fashion? My sequence would be:

  1. Noise reduction
  2. Normalization
  3. Split stereo to mono
  4. Declicker first pass for pops, second pass for clicks
  5. DeEsser

When I split the stereo to mono, can I delete either track, regardless? (Are differences very subtle?) Or, should I listen to both mono tracks to see if there is one I like better, and discard the other?

If I had not already done 1, 2, is there a different order? That is, if I were starting a new project, would the order be different? (Probably the split first, but I’m not sure about the rest.)

JB

Your stereo-tracks only very slightly different : if you “remove center” from a few seconds of a stereo track as an experiment, the result is very quiet, that means the left & right are almost identical. So one track is as good as the other : delete either.

Yes split first will halve the workload, reducing the processing time.

Audacity’s native normalization tool is based on peak-values of the waveform. If you want consistency of volume across different recordings, use normalization based on the average (RMS) value. A plugin for RMS normalization is here.

The default settings on the DeEsser plugin may not be optimal. Here is my suggestion …
suggested DeEsser settings.png
Other free DeEsser plugins are available which are easier to use, e.g. SpitFish. (Windows computers only).

Trebor,
This is great. And now that you’ve mentioned normalizing multiple files, let me ask you this because it’s another issue for me. The solution I found online, which I tested and seemed to work, was to convert all files to mp3 (which they need to be for my project), import them all, apply normalization (the regular one in the menu with no changes to settings), and then I chose export individually. I used four files of differing volume; the results seemed consistent.

Do you think using the RMS tool would be better?

JB

was to convert all files to mp3 (which they need to be for my project), import them all, apply normalization (the regular one in the menu with no changes to settings),

No! MP3 is lossy compression. If you want MP3, compress to MP3 ONCE as the final step. If you are saving intermediate files, export to WAV.

When you open an MP3 it gets decompressed. If you re-export again to MP3 you are going through a 2nd generation of lossy compression. The “damage” does accumulate and the quality loss can be4come audible. MP3 can be very good and can often sound identical to the uncompressed original, but keep in mind that it is a lossy format. Its a distribution format, and should be avoided during audio production (if possible).

Do you think using the RMS tool would be better?

If you want “equal loudness”, RMS is better, but it’s best to do it by ear unless you have hundreds or thousands of files and that’s impossible. (Or, unless you’re in a hurry and you want to take a shortcut. :wink: )

I’d normalize first and if the perceived volumes don’t match, you can make some further adjustments. After normalizing all of the files you can’t boost the quieter files without clipping, so choose the quietest-sounding file as your reference and adjust the louder tracks down to match (by ear or from the RMS levels).

Oh! Well, then, I’ll scratch that method!

I actually did a few of them by ear, but it was super-tedious because I had trouble knowing the increments to use to get louder (I was matching to the loudest as it was the volume I wanted.) I have about 25 files. Many of them are at a good volume level already, so it’s a question of matching the others to them. I was just doing that with the .aup files. I suppose after I do a few, I’ll get the hang of it.

On the other hand, I’m seriously thinking of starting all over and doing all 21 at the same time. Then I only have to balance the four done by other people.
At this point it might take less time…Sigh…

That explains how you managed to create a slightly-stereo track from a single mic : mp3 format is an approximation, the two stereo channels created are slightly different approximations of the same original mono recording.
Only use mp3 format for the final finished product you send to the customer, (if they insist on mp3).
Use WAV (or FLAC) formats when creating your project, as they are precise copies rather than approximations.

RMS normalization is more reliable way of obtaining consistent volume than Audacity’s built-in (native) normalize tool.

You can manually adjust volume of anomalous sections using Audacity’s envelope tool

Trebor,
I’ve been working on my files (I’m technically on vacation, but not quite… at least the pace is slower). I’ve had success with many of them, but there are one or two sticklers.

In one of them I can actually see the pop as it’s part of a silent spot, so I can see the effect the Declicker has. It reduces greatly with the first pass and settings. Barely at all with the second. In two of my files, the pop is still faint but it’s there. I ran the first pass (your settings way above) twice, and the second one twice. (Of course for the one in silence, I can just snip it out, but I was interested in seeing the ‘visual’ effect and matching it to what I heard.)

Is there a slight modification I can make to the first pass settings that will make that pop even quieter?

I doubt that my listeners will notice it when there is voice with it. A few people I’ve asked don’t hear it until I point it out… but I know it’s there! If there’s not a simple adjustment to the settings, then I’ll just leave as is, or re-record the few that offend me :slight_smile:

Thanks so much–this has been a real help, especially saving the settings as presets.

JB

Rather than running DeClickers twice, run once, but with the “Number of passes” parameter increased




Reducing the “sensitivity threshold” parameter will remove more, but as it is lowered, at some point it will remove too much and cause distortion of the speech, ( I wouldn’t set “sensitivity threshold” lower than 3 ).

The DeClickers won’t catch everything. The pops & clicks they miss can be removed manually using Audacity’s spectral edit tools

You may have to increase the Audacity spectrogram resolution from default setting of 256, to 1024, or higher to see the bassy pops.