Render vocals indecipherable; maintain human speech quality

[AlwaysSunny]

I should mention this is for a radio or walkie talkie effect, since I imagine that makes a big difference as far as options. I have a nice clean "phone quality" recording, and achieved a decent radio effect via equalization and a few subtle tweaks. I'm not happy with it though, and I'd like to be able to garble the track so excessively that individual words cannot be deciphered. (This track is just for ambiance anyhow.)

I know I can amplify the sample until a sufficient amount of data is clipped, but this has the unintended side effect of creating a nerve-grating quality. I'm wondering if there's a better way. After a good bit of experimentation, I thought I'd just ask.

I really appreciate your time, thanks for reading.

[steve]

Try this "reverso" effect.
It takes short sections of the sound and reverses them.
http://forum.audacityteam.org/viewtopic ... 65#p160165

Instructions for installing the effect: http://wiki.audacityteam.org/wiki/Downl ... g_Plug-ins

[AlwaysSunny]

Wowzers, thank you so much for this quick reply!

Really interesting effect you've got there. Are you the author, or was that a different Steve? Thanks for reminding me of that repository; now I've got some new toys!

After a bit of playing around with this, I'm still not totally clear on what it's doing. It seems to reverse, shuffle, and bleed data around a bit. Quite interesting indeed. Through a few iterations and the native Reverse effect, I managed to dial in something darn near passable.

To my ear though, regardless of settings or iterations, it creates enough of an inhuman quality that I'm shy about saying it satisfies both halves of my desired outcome (indecipherable but human). I fully intend to continue experimenting with it, though I'm also eager to hear any other thoughts; perhaps ways of clipping a sufficient amount of data without creating ear-sandpaper? I'm very much an amateur; perhaps there's an obvious answer you might assume I already know about?

A great big thank-you for your time,

[steve]

Really interesting effect you've got there. Are you the author, or was that a different Steve? Thanks for reminding me of that repository; now I've got some new toys!

Yes, I'm the author.
Please note that plugins in that section of the forum are "unofficial" and mostly "experimental", though most of them should work pretty well. It is a big help for "promoting" plugins to the more official wiki page if people leave feedback in the topics where they have downloaded plugins.

The "more official" plugin page is here: http://wiki.audacityteam.org/wiki/Downl ... st_Plugins but please note that some of those plugins are very old. They should all work (please let us know about any that don't), but some may be a bit, (how to describe...), "primitive".

To my ear though, regardless of settings or iterations, it creates enough of an inhuman quality that I'm shy about saying it satisfies both halves of my desired outcome (indecipherable but human). I fully intend to continue experimenting with it, though I'm also eager to hear any other thoughts; perhaps ways of clipping a sufficient amount of data without creating ear-sandpaper?

Does it matter if you can make out individual words? I've got another experimental plugin somewhere that chops the audio into bits and shuffles the bits around. I'm not sure how complete that plugin is - I'll have a look to see if I can find it, and if it looks usable I'll post it.

[AlwaysSunny]

I hope I haven't coerced you into going above and beyond here, but I certainly appreciate any such efforts. Please don't go too far out of your way, but the effect you're describing does sound potentially fun and useful; perhaps for others' purposes also.

My audio sample is an air-traffic-control recording, and because I'm not intimately familiar with laws regarding duplication, I'm anxious about leaving the words distinguishable. Besides this, distinguishable words are undesirable from an aesthetic standpoint in my scenario. (key mood-setting ambiance for a game). So words are bad; but I want to maintain the impression that a human being is speaking a language with human vocal cords, over a really, really lousy radio.

I can come mighty close by equalizing, amplifying, pass-filters, and normalization, but the final effect seems much too harsh. (I may very well be asking for something nearly impossible for all I know?)

Very much obliged,

[Trebor]

... perhaps ways of clipping a sufficient amount of data without creating ear-sandpaper? I'm very much an amateur; perhaps there's an obvious answer you might assume I already know about?

Clipping creates higher frequencies , which are the "sandpaper" effect .
You can filter them out (after clipping) with this equalization if you want a walkie-talkie effect ...

suggested equalization.png (18.04 KiB) Viewed 978 times

[AlwaysSunny]

@Trebor ------------
I'm a little embarrassed to admit how well this satisfies my request as written. Perhaps I neglected to try a similar equalization after amplification; it sure feels as though I've tried everything!

This produced an effect that much closer to what I envision, and definitely nullifies language recognition without sounding too ugly. I'm still struggling to reproduce the finer qualities I hear in my imagination, but I'm certainly in the right neighborhood.

@Steve ------------
I'm still quite curious to try out the aforementioned plugin, but again, don't feel obligated. I know what a bummer it can be to dig back into old code. Interestingly, in-game, I have one ten minute radio_chatter.ogg I play short snippets from at random timecodes; so I could actually achieve the "chopped n' shuffled" effect at runtime if I so desired.

[steve]

I'm still quite curious to try out the aforementioned plugin

Found it
http://forum.audacityteam.org/viewtopic ... 17#p269317