You shouldn’t need to go as low as -30 dB. Use a compressor and/or limiter and/or the envelope tool (Envelope Tool - Audacity Manual) to bring the audio to a more even level before you normalize the RMS.
It depends on when you hit clipping.
If you hit clipping during recording, then you are toast.
If you hit clipping on exporting, then you are toast (unless you export as 32-bit float, which is not generally recommended because few programs support 32-bit float audio)
If you hit clipping while working in Audacity, then you still have fresh bread (not toast), provided that your track is 32-bit float format (default).
One of the most useful features of 32-bit float format is that it doesn’t stop at 0 dB, but can go on for hundreds of dB above the normal clipping level, so all you need to do to bring the audio back into the “valid” range (below 0 dB) is to apply the Amplify (or Normalize) effect with default settings.
As a demonstration:
Import some audio. Ensure that the track info panel says “32-bit float”.
Apply the Amplify effect with default settings (the peak level is now 0 dB)
Apply the Amplify effect with +30 dB and the “allow clipping” option enabled. The peak level is now +30 dB and will sound very loud and distorted.
Apply the Amplify effect with default settings and “allow clipping” disabled, The peak level is now 0 dB and there is no damage.
If you hit clipping while working in Audacity, then you still have fresh bread
That’s why it is recommended you announce live or present with occasional peaks at -10dB to -6dB, so you don’t unintentionally overload before you ever have the chance to correct the errors.
Once you’re inside Audacity, assuming you left everything alone, there is no overload. Apply the rumble filter to get rid of low frequency damage, apply RMS-Normalize to put the loudness where you want it and Limiter to gently squash the peaks. The goal of the suite is to achieve the desired technical values and still sound perfectly like you.
Also please note the odd limiting value of -3.5dB in the suite is designed so the show peaks are lower than -3dB after you export to MP3 . MP3 can create peak problems. It’s good to allow for that any time you get into and out of MP3.
Which brings us to: Never do production in MP3. Do everything in Perfect quality WAV and convert to MP3 as a last step if you need to.
Given you have multiple performers, my opinion is you should try Chris’s Compressor. The poster child for Chris is the Car Talk radio show where one performer presents loudly and has a nuclear laugh and the other mumbles in his beer. Pretty much your original complaint.
Don’t cut the end off the final edit until you run Chris. Leave the goodbys and post production trash on there to make Chris happy. After processing, then cut it off.
I did a graphic of the compression effect. No Processing (raw capture), Default Chris at 0.5 and my version at 0.77. The waveforms get denser and the difference between high peaks and overall average starts going away.
Chris designed this processor so he could listen to opera in his noisy car. So everything between tutti orchestra and one violin in the south forty comes out about the same.
This is enormous news for me, thank you so much. I had never made that connection. This changes a lot of things.
Have been trying out Chris’ compressor too, and that seems to be really helpful. This thread is going to save me a lot of time and effort, I think, thank you all so much.
It’s possible to create special purpose sound files and other work for unusual jobs, but keep your eye surgically on the goal. If you ever expect anybody outside of your cohort to play the files, it’s best if they’re in a common format.
Time offsets can create problems, too. “Oh, wait. I think the only machine that can play my Frammis Format files is out in the garage.”
Thank you for the RMS Normalize plug-in. It’s exactly what I was looking for.
I use Audacity to record the voice over (narration) for videos that I produce in Microsoft PowerPoint. (It’s okay, you can laugh .)
I’ve used PowerPoint for many years to develop slides for “live” presentations. More recently, I’ve been using PowerPoint to produce videos. I export the presentations from PowerPoint, with timed effects and voice over, to MP4 format. So far, the slide transitions and animation effects in PowerPoint meet my needs. And I like the modularity of this approach. It’s very easy to insert new “scenes” (slides) and combine or split videos.
I was already using Audacity to record the narration before I found the RMS Normalize plug-in, but I had a problem. My “modular” approach in PowerPoint meant exporting the narration for each slide to a separate file (I export to WAV, 32-bit float PCM). I was applying a chain that included Noise Reduction and Normalize; that is, peak normalization. There were jarring differences in loudness between audio clips.
The solution was to replace Normalize with RMS Normalize. Thanks again!
Are there any issues with this plug-in and Audacity 2.2.1? On my Mac I’ve copied the plugin to “Library–> Application Support → audacity (and also in libs just for kicks)” and am trying to use the “Add/Remove Plug-ins” tool, but no luck.
Thanks for any assistance you may be able to offer.
I don’t quite understand how to install these plug-ins (both the rms and acx check). I’ve downloaded them both in firefox, but neither show up in my attempts to add the plugins and enable them following Audacity’s instructions.
So here’s my question. Complete amateur here… I normally use a plugin called Voxformer by Voxengo. It’s magic and does wondrous things to my audio for my books. My normal procedure is to run Normalize, Voxformer and then Limiter. How would I use the RMS Normalize in this lineup?
Follow the formula exactly, please. Post the raw test here.
Now run the test clip through your normal Voxformer process and write down or screen grab the settings. Post the processed sound clip and the settings.
We note Voxformer is pay-to-play software. How much was your license? A simple, one-user license seems to be $395 usd.
.)