Punch/copy/paste and the order of effects for audiobook

I have an unpopular, fuzzy rule that if you make it into your second week of trying to force your computer to record your voice, you should probably stop recording on the computer.

I set up briefly for a sound test just to see how it would work. And it does. This is Apple Voice Memo.


Colonel Mustard, In the Garage, with an iPhone.

There are some file transfer difficulties with this method, and it’s not a complete sound isolation from the neighborhood. But if all your other recording tests are terrible, this does work.


Koz

Right then.

I developed the correction a bit. In my opinion, it sounds like you sitting across the table from me. The sound sample is the new correction followed by a bit of uncorrected. See what you think. This correction is not difficult and it can be applied anywhere in the process—but only once. So bookkeeping is important (as in most editing).


This is the simple list of sound tones and what I did to them (more later).

Gill-OCurve.txt (96 Bytes)
There is a method of producing a preset that you apply to the effect and I’m developing that.

As we go.

Depending on how you’re listening, you may like the latest “fix.” Do post back.

Koz

Progress ! :smiley:

I’ve tried the oblique placement and yes, it does hold advantages. The mouth noises are a bit better, and there’s no risk with plosives - which was not a major issue, but an occasional irritation. It also means I can have the music stand closer and see more at a glance. I’ve re-thought the position of the computer and that too is an ergonomic gain.
Tips on lighting welcome - I worked today with a little chink in the curtains (I’d moved from paper to iPAD so as not to have the light on - but take your point, that’s just swapping one source of interference for another.) The window is behind the mic.

Your correction sounds really great to me. WOW. :smiley: Thank you so much. So next up is to learn how to do it.

RE your iPhone set-up - that’s impressive, but unless my recording tests really are that terrible, I’ don’t think I’m nimble enough.

I have set up a spreadsheet to keep tabs on it all, promise!

Your correction sounds really great to me. WOW.

Excellent. The design is to suppress the irritating tones without sounding muffled like I threw a duvet over your head. That correction, too, will help suppress mouth noises.

I did think of a process “problem.” Filter Curve EQ is used twice. Once to get rid of the “gritty” sound and then again as part of Audiobook Mastering—with different settings.

That’s just asking for an accident, so I dug up a project I designed about a year ago. Audiobook Mastering doesn’t have to be three individual steps. It’s possible to create a Macro document that automatically runs all three one right after the other. You just point at the performance, start the Macro, and stand back.

We also haven’t settled the question of edit order—the original question.

unless my recording tests really are that terrible

You are way further ahead than many first time users. We would regard Michael who wanted to read audiobooks from his apartment off La Brea in Hollywood (the actual geographic location, not the metaphor).

We succeeded and he’s a successful commercial reader, but it took 39 forum chapters and over a year.

I think that’s the record.

As we go.

Koz

A little research. I don’t think you ever told us which Audacity you’re using. Some of the advanced tricks don’t work in the earlier Audacity versions.

Koz

Let’s set up the gritty filter.

Open Audacity and open something on the timeline. Doesn’t matter what. Generate > Noise, 1-minute > OK.

Screen Shot 2022-07-03 at 4.44.06 PM.png
Effect > Filter Curve EQ…

Pull the EQ window wide so you can read the numbers on the left and bottom. Note there are sliders on the right that affect value detail and visibility.

Flat-EQ.png
If you have used this tool before it will have the last used pattern, not flat. Press “Flatten.”

You can click on that flat green line and move it up and down. If you click on two different places and push, you can get a bend, etc.

Click once on the green line above 1000Hz and leave it at 0dB.

Click above 2500Hz and drag downward to -9dB.

Click above 4500Hz and leave it at -9dB.

Click above 8000Hz and drag up to 0dB.

This is that recipe from earlier.
Gill-O Crisp Compensation
1000Hz 0dB (Anchor Point)
2500Hz -9dB
4500Hz -9dB
8000Hz 0dB

You should get a pattern that looks like this.

CrispCompensation-EQ.png
You can save this pattern so you only have to draw it once.

Manage > Save Preset. Call it something you’ll remember > OK.

From here on, any time you want that pattern and method of smoothing the sound, Effect > Filter Curve EQ > Manage > User Presets > Choose your correction. That pattern will appear. OK will apply it.


Those numbers along the bottom represent pitch tones. Low pitch (rumble, thunder) on the left and high pitch (frying bacon, air leaking) on the right. Even if you never met this graph in your life, you know what 3000Hz to 4000Hz is. That’s baby screaming on a jet.

Next is the fancy Mastering tools. Do let us know which Audacity version you have.

Koz

The window is behind the mic.

I couldn’t put that together until just now. You have heavy drapes at the windows to keep street noises out?

Your microphone claims to be directional so anything arriving at the rear is greatly reduced in volume.


You needed to crack the drapes a bit to see your paper script? There are some interesting problems with artificial lighting. Classic tungsten light bulbs are perfectly acceptable, but don’t dim them. They start singing if you do that. eeeeeeeeeeeeee.

Fluorescent tubes were traditionally terrible. They featured very good efficiency and a quiet buzzing sound.

We had a discussion at work once as to who found the quietest compact fluorescent light bulbs. Many of them made annoying noises. That was the fight between good efficiency, brightness, size, and cost.

The newer light emitting diode LED bulbs seem to work OK.

You can do a real-world noise test. Turn your lights on as you wish to use them and start a recording with the microphone turned all the way up. You can do this in real time with your headphones cranked up.

Move the microphone close to the lamp and point it at the bulb. Wave it around a bit. you can work up a process based on the results. Even if the bulb is gently noisy, if it only picks up with the microphone right on top of the bulb. It might be OK.

I have several room noise events. I suffered from muffled hum for a long time and just couldn’t find the cause. It’s not apparent in real life. I had a special location where I could put the microphone and stand and it mostly didn’t pick up the hum.

I finally did that wave the microphone around thing (picture the guy with the metal detector at the beach) and the hum turned out to be coming from my musical keyboard bass cabinet. It didn’t go off when I turned it off. mmmmmmmmm Surprise.

Two other noisemakers are the internet data interface. That one screams quietly, and the Ikea wall clock.

Tick
Tick
Tick

Koz

Can you tell if your computer is on just by listening? Not good news.

Koz

Thanks again!

I’m on Audacity 3.1. Thanks for your instructions on the gritty filter which I’ll tackle later today. You call it a gritty filter, to me it’s the aural equivalent of a facelift. :laughing: I had thought of it as a final polish but in fact it should be done before the mastering, yes?

Thanks to other posts on this forum I had set up a macro for the basic mastering and am neurotically logging the numbers on ACX check to make sure they’re consistent - although why wouldn’t they be? Well the main thing that scares me is stuff that I don’t know, and there’s a lot I don’t know in this area. Also human error.

The mic faces away from the window, which looks on to garden,+ neighbours gardens. Any noise from that is intermittent - neighbours kids, helicopters etc - and to be honest the directional mic hasn’t picked up any of these things or at least rendered them in audible form, even though I can hear dogs, rainstorms though the big fat Audiotechnica headphones when it’s happening. I usually stop because noise distracts me anyway. I always make sure heating, washing machine etc are off (yes I’ll have to get this done before winter sets in!! :unamused: )

The computer isn’t noisy, I was just trying to be careful. I always disable the internet when recording.

Thanks for your suggestions re lighting, I’ll experiment with that too. Many thanks!

‘When to edit corrections’ is always the mystery. You can apply the gritty filter right after you finish the first pass reading—when you’re not likely to need any new reading on that chapter.

Then it gets messy. Editing is handy after Mastering because the blue waves are strong and tall and it’s easy to see defects, ticks, and pops. Also, the chapter has no rumble to distract from editing.

There is a trick with Mastering. Step 2, Loudness Normalization and Step 3 Soft Limiting go to completion. If they’re not needed, they don’t do anything. You can apply them multiple times, but Step 1, Filter Curve EQ (rumble filter) can only be applied once.

So if you get done patching and editing and the chapter will no longer quite pass ACX-Check for RMS or Peak, you can apply Loudness Normalization and Soft Limiting manually. You can’t use the Macro because that automatically applies all three steps.

Getting rid of mouth noises shouldn’t cause ACX-Check to fail, but stranger things have happened.

Did you notice in that sound file I prepared the strident ssssss in the background was quieter in the filtered segment? It may turn out that you may not need Noise Reduction at all after Oblique Microphone Positioning and the Gritty Filter. Or to push the noise value into “safe” territory, Noise Reduction of the Beast (6, 6, 6).

6, 6, 6 is a little magic because nobody can tell you used it. It leaves little or no sound damage.

“Overprocessing” is a common ACX failure. This can be experienced by a performer who believed the microphone manufacturer’s assurance that you can set up on the kitchen table and clean up the resulting errors later in software. Design of software filters is a juggling act. In general, the faster they work, the less desirable the results. You can design many corrections and filters to produce almost perfect work if you’re willing to wait the days of processing it might take to do it.

Everything is a compromise. This is why the closer you can get to a perfect, quiet studio reading, the better.

Koz

Thanks, I’ve set up the gritty filter and done a test with a 7 min narrated file . Noise reduction 6:6:6 got it from -57.51 to -63.51dB.

I agree that once you’ve done the first pass, it is much easier to edit after mastering.
Which means I’ve got to get the mastering right, from the start, and stick with the same recipe for each file.

I get that it’s not a good idea to over-process. The grey area for me is, what counts as processing, and what counts as editing?

Cut and paste = editing, de-clicker = processing ?

My impulse is to squash every little irritation. Every gap between speech / sentences has to be right - comic timing don’t you know - and editing is going to be slow with or without mouth noise… I could so over-do this… :unamused:

But I am strong. :laughing:
All the best,
Gill

de-clicker = processing ?

Only in the sense that it’s extraordinarily difficult to get de-clicker to sound right—or not sound like you did anything, which is the actual goal.

what counts as processing, and what counts as editing?

Say you make a fluff or mistake, pause to re-announce the ‘whole sentence’ correctly and then on to the rest of the chapter. Later, in editing, you cut the bad sentence in favor of the corrected one. There’s a New User mistake where you try to only fix ‘the word’. Bad idea. When you listen to the section with the correction, you can’t tell what you did. That’s editing.

Obvious processing is trying to rescue an insanely noisy chapter by stiff noise reduction and resulting in a presentation that sounds faintly like speaking into a wine glass or milk jug. There are others. Bad application of noise gate which cuts off beginnings and endings of words and word quality that sounds different from the spaces between words. “Talking air conditioners” is an example.

Noise reduction 6:6:6 got it from -57.51 to -63.51dB.

In my opinion that’s too close. Again, according to me, you have to clear the ACX specification by at least 6dB. You can stiffen up the noise reduction to as high as 9:6:6, but that’s pushing it. I’d be interested in what kind of noise you have.


ACX’s explicit desire is a presentation that sounds natural with no, as in zero, distractions. My version of that is making you sound like you’re telling me a fascinating story over cups of tea. Correctly done, my tea gets cold because I’m paying attention to you.

There’s a Youtube presenter I can’t watch. His video is perfect, but his voice sounds like recording in a bathroom. There’s just something about an authoritative, professional voice coming from over there by the tub.


There is a New and Improved format. ACX used to not like Radio Theater. Multiple voices, music, special effects, accents, etc. Just you reading the book, thanks.

I understand they’re OK with radio theater, now. I saw an Audible company promotion which said that. That decision turns a ticklish voice editing task into a hellacious multi-format production job.

Good luck.

Every gap between speech / sentences has to be right - comic timing don’t you know

I do know. I was in a Cast and Crew screening of a comedy movie. It was…ok. When I was leaving the theater, the people around me (remember, professionals) talked about anything other than the movie. “Did you hear there’s a backup on the 405 freeway?”

I saw the “real” movie in a theater later, after careful re-editing, and it was hold-your sides and spill your popcorn funny.

You know what the rules are and what it’s supposed to sound like. There’s another New User condition. Nobody likes their own voice. The first time they hear a recording they’re horrified. “Do I really sound like that?!?”

I think you’re going to fall into that bit where you get to the end of the book a relaxed, seasoned professional and go back and re-read the first couple of chapters.

Koz

I would have no trouble listening to a story in that voice.

Koz

For some reason both Punch Copy and Punch Paste did not appear anywhere after being enabled. I uninstalled and reinstalled Audacity, then reinstalled Punch Copy and Punch Paste. This time only Punch Paste appeared on the Tools menu, but doesn’t work withou Punch Copy. Any ideas how to fix it?

Thank you all for this great conversation - I’m just getting started and it all applies!

did not appear anywhere after being enabled.

Which Audacity are you using? That may be one of the tools that only works in Audacity 3 and higher.

You’re also talking to the presenter whose voice is so bad I never got good at the patching tools. We can wait for one of the other elves.

Koz

I’m using the latest available download, version 3.1.3

When I originally tried to download the Punch Copy, there were issues with it not downloading - was wondering if they incorporated it the main program, but I don’t see it anywhere.

I got it fixed!

Turns out that the one I originally downloaded was named PunchCopy.ny and was from around 2013. In the script it was named Punch Copy/Paste.

I found a 2021 version in the forums called PunchCopyPaste.ny and it installed and showed up perfectly :slight_smile:

I’ll see if the documentation elves know about that.

Koz

There are at least 3 versions of Punch Copy/Paste on the forum.

The oldest versions are clearly marked as “obsolete” and should not be used.

The two versions that are still current can be found here:
https://forum.audacityteam.org/t/punch-copy-paste/28906/1
and here:
https://forum.audacityteam.org/t/punch-copy-paste-problem/64114/4


The older version is in this forum post: https://forum.audacityteam.org/t/punch-copy-paste/28906/1
There are two current (and two obsolete) plug-ins:

  1. Current Version: PunchCopy.ny: Is the original “Punch Copy/Paste” plug-in. When installed it appears in Audacity’s Effects menu.
    It may be used to copy, paste or clear it’s clipboard.
    This version is not designed to support batch processing.
  2. Current Version: PunchPaste.ny: Is the original “Punch Paste” plug-in. When installed it appears in Audacity’s Effects menu.
    This plug-in only works with Current Version: PunchCopy.ny, and is intended to be bound to a keyboard shortcut so that a copied selection can be quickly pasted many times.

Comprehensive documentation for this version can be found on the download page.



The newest version (at time of writing) is on this page: https://forum.audacityteam.org/t/clicking-enter-as-part-of-a-macro/61687/14
This version was written to support use macros and batch processing. It is NOT COMPATIBLE with other versions.
When installed it appears in Audacity’s Tools menu.
Usage is very similar to the other version except that it has an option to copy silently (no message after copying).



I was rather hoping that Audacity would eventually get a built-in “punch copy/paste” like feature rather than requiring a plug-in, but as I’m no longer doing C++ development that seems unlikely. Nevertheless I’ll post it as a suggested “feature request” on GitHub.

Posted here: Punch Copy/Paste · Issue #3194 · audacity/audacity · GitHub

If you think this would be a good addition to Audacity, please feel free to say so on the GitHub page (requires a free GitHub account).