View Full Version : Disguising voices in FCS?


Adam Bray
December 13th, 2007, 11:34 PM
How do I disguise voices? I have an interview clip and am looking to alter the audio to disguise the voice.

You have seen this done before on TV when people like informants an undercover police are being intetviewed.

David W. Jones
December 14th, 2007, 12:03 PM
It always amazes me when I see badly concealed informants on TV,
knowing that the Bad Guys will just hire a pro to simply reverse the pitch shifted voice and mosaic face.

Kelly OHara
December 14th, 2007, 02:57 PM
It always amazes me when I see badly concealed informants on TV,
knowing that the Bad Guys will just hire a pro to simply reverse the pitch shifted voice and mosaic face.

FYI: Voice pitch shifting doesn't destroy a lot of information so can be easily reversed, but still a bit tricky to find the right pitch return. It's usually pacing and inflection that help a person to identify someone vocally.

However, the mosaic effect or heavy blurring is massively lossy... it normalizes tens or hundreds of pixels into a single value so will not be reconstructible back to a usable value. To regenerate a face from a mosaic or blur is like saying you can re-create any of your neighbour's answers on the census based off of the final numbers.

That being said, if there was a thumbnail or EXIF information with a still, you could regenerate it from that.

If you're inferring the reversing of the swirl used to identify that pedophile can be applied to any obscuring technique, the swirl doesn't destroy nearly as much information, it just relocates it... so you just have to figure out the right method that was used to create it and reverse it... (it's the same as pitch shifting a voice, the information is there, just relocated)