In early September we posted about Oona Räisänen’s deinvert which is a tool that can be used to unscramble voice audio that has had voice inversion scrambling applied to it. Voice inversion works by scrambling the voice frequencies so that a simple eavesdropper will have trouble listening in. A special descrambling radio is required to listen in. This provides very little real security, but may be enough to stop people with cheap scanners from listening in. Oona’s deinvert tool allows us to take a scrambled audio sample recorded with an RTL-SDR or any other radio and decramble the inversion.
In her latest blog post Oona explains how her deinvert software works and how it can also be used to decode the more difficult split-band inversion technique. She also writes that at the default quality level, the deinvert software is fast enough to run in real time on a Raspberry Pi 1.
Voice inversion scrambling is a simple and old security method used on analog radios to try and obscure conversations from being listened in on by people with scanners. It works simply by by moving the low frequencies higher and the high frequencies lower, or in other words inverting the audio. A descrambler is then required to recover the true audio, otherwise you will only hear garbled audio. Voice inversion provides little real security, as it is very simply to descramble, and many scanner radios already have descrambling features built in. These days most secure communications are digital and encrypted, but voice inversion scrambling is still available on many analog radios, and could still be in use by some users looking for protection against casual eavesdroppers.
Oona Räisänen (aka windytan) has recently released a simple program called ‘deinvert’ over on GitHub. This program is a descrambler that reads in a scrambled wav file and outputs a descrambled audio file. The audio file could be easily recorded with an RTL-SDR and rtl_fm, or a similar SDR.