Step 1: Unsurprisingly, deep or male vocals work best for this Aphex Twin-style effect. First, resample your original vocal - we’re using this one:
Lower it by three or four semitones in pitch, keeping the timing the same. Lower the volume on the original sample slightly and resample the two together. This should produce a thick, cavernous sound.
Step 2: Place a distortion plug-in such as Audio Damage’s free FuzzPlus 2 on the new recording. Set the Fuzz and Tone to be moderate, but not overpowering. Depending on the pitch of your original recording, you may need to experiment to get it right.
White noise generator and vocoder
Step 3: Discard everything apart from the resampled audio. Now add in a MIDI channel and a new audio channel. Add a synth with a white noise generator to the MIDI channel and a vocoder to the new audio track.
Step 4: Set the synth to produce white noise only and draw in one long MIDI note, the length of the vocal sample. Route the MIDI to the vocoder channel as the carrier. Route the vocal to the vocoder channel as the modulator.
Step 5: Adding EQ can make the vocal clearer. If your DAW has formant presets, try these as they emphasise the natural throat sounds. If the result is too noisy, lower the distortion earlier in the signal flow. Finally, some very gentle delay on the output can add an atmospheric finishing touch. Here’s the end result: