Help needed cleaning screeches/peaks from voice recording

I used my phone to record a convo. It recorded a lot of screeches or peaks in the voice (I dunno the exact term, see waveform). The phone was in my pocket, so not really close to the mouth.
Screenshot (16).png
Here is the sample file from the selected part.

I tried compression, but it just compresses it as a whole, and the voice is still the same as before, screeches are still there, just a bit less loud. I tried EZ-Patch. It helps to blend these peaks with the nearby sound. Thats great for music, but it distorts the voice even more.

Any ideas how to proceed?

Audacity’s native tools won’t make much difference to this severe case of clipping distortion.

You’ll need $oftware with automated de-click and de-crackle , even then it’s an extreme-case and parts may be unfixable no matter how much is spent on software.

holy shit, that was awesome!

can u share how you actually declipped that? doesnt matter if it isnt audacity. Now that i understand clipping, i know that the data has been lost, because i set too much mic gain on my phone, not knowing the end result would be like this. and its impossible to get back that data; softwares can only predict to reconstruct it. But i just want to get rid of the screeches as they are really irritating, and to automate the reconstruction as far as possible, as the audio is an hour long, and I cant really edit the waveforms manually.

Here’s a method using Audacity only, ( you’ll need to add the “median” plugin) …
Repair clipping using 'median' filter#.gif
The text you need to paste in the Nyquist Prompt is …

(multichan-expand #'slope s)

(“Nyquist Prompt” is in Audacity’s effect menu).