I'm trying to find a way to process a speaking voice, to give it a hollow, breathy, possibly reedy quality. I would like it to sound slightly transformed - uncanny - but subtle enough not to sound dehumanised. The best reference for this would be the singing in this track:

I think I can hear some autotune at work here, and I tried playing about with Melodyne but as this is a speaking voice it isn't noticeably altered unless you really push it. My second thought was, well I'm not interested in tuning per se, but rather the artefacts of tuning. So I turned to formant shifting. I got somewhat close by formant shifting down, and compensating by pitch shifting up - but it's perhaps too much of a transformation, or produces too many side effects.

I was combining this formant shifting with some compression in the mids, which also helped. However, I don't think I could rely solely on EQ or multiband dynamics. That is useful for subtle adjustments, and perhaps in combination, but on its own to perform a transformation like this it would just start sounding tinny or similar.

The thought I've come to rest on just now is that I would like to suppress tonal/periodic components while keeping or enhancing noisy components. I know I could measure noisiness via zero-crossings or similar, but can't think how to proceed from there.

I use Reaper, Ableton, and Max/MSP. I'd be happy for solutions that point me to a plugin, a technique, or an approach for patching something.

I could give you some ideas for this vocal mix.

Record different vocal takes and chop them in a word-by-word style. I do hear some chopping which produce small inhuman changes between words.

There's certainly a machine working here and by machine i mean 2-3 channels with different plugins interacting with each other based on dynamics/frequencies/effects/phase.

The silent-(up-front)-dry / loud-wet effect: The vocal is somewhat dry when not singing loud (0:20 and on) and wet with doubler/chorus and space fx when louder (2:03 and on). To achieve that effect , copy a vocal to another track , apply all the FX in the first vocal , keep the copy dry and smash it with an optical compressor to be more musical. Mix these 2 tracks and let track A (wet & uncompressed) surpass track B(dry & compressed) in volume when the singer is loud, it takes some time to master , but it'll give you that dry/wet effect.

The bright layer. The vocal has much musical content in the mids/lower mids but there's a constant whisper and bright layer going on throughout the mix. One cool way you can achieve that is by utilizing the dolby A trick and using only the D band. I'm sure you'll be able to find some dolby A enhancer plugins out there, there's also this SOS article on how you can go on about creating one (https://www.soundonsound.com/techniques/bright-idea). What this will give you , is essentially a bright layer without being just EQ:ed and somewhat irritating.

Phase. A cool way to produce weird effects is to play with phase. I would definitely try this to produce a vocal like that.

A simple trick is to take a phaser which has a static option and a feedback option and play with it , find a spot that sounds cool and close to your needs both in frequencies and amount. A Wiki on how a Phaser works: https://en.wikipedia.org/wiki/Phaser_(effect) .

Another trick , is to create a machine with a dry and a dynamically processed track (let that be expansion/compression/reverse-compression or whatever you like). Start by copying the vocal to another track and flip the phase so the VU meters jump but there's no sound (cancellation). Now keep in mind that whenever the 2 tracks come close to amplitude the properties they have in common will be cancelled out im saying this because our tracks wont be identical!. So let's say you compress track B by a good amount, now , whenever the track A dips in amplitude and reaches the track B levels there will be cancellation same goes with expansion or even EQing / distorting / saturating. The more effects you chain the less profound will the cancellation be , because you're altering the signal completely.Also this can be done with fader riding!

If you can record the vocal, you can also use 2 mics, flip the phase on one mic, and play with the distances until you have a cool effect on the vocal, this is a very powerful technique when used correctly and a very hard one!

Re-amp If you already have recorded the vocal , and you want to experiment with mic phase effects you might as well re-amp it!. Simply play it through a speaker and record it again with another technique, you might even play it through a big fender twin reverb and record the room or the amp , or the back of the amp or whatever sounds cool!

Also the SoundToys AlterBoy plugin can create some cool effects! :D

Most of these are experimental techniques focused towards what you originally posted. Whenever i want to make a cool/unique sound these are some tools i use.

Good luck :)

  • This is a wealth of great ideas! I think the bright layer was what I was particularly in search of, so definitely intrigued by the Dolby A trick. Also you're probably onto something when you suggest playing with phase. Some of what I was hearing as formant shifting is actually probably down to phasing as well. Thanks so much! – Igid Apr 26 at 14:21
  • This is a wealth of great ideas! I think the bright layer was what I was particularly in search of, so definitely intrigued by the Dolby A trick. Also you're probably onto something when you suggest playing with phase. Some of what I was hearing as formant shifting is actually probably down to phasing as well. Thanks so much! – Igid Apr 26 at 14:21
  • Glad you find these ideas interesting! Don't get disheartened if you dont nail the phase at first, it's a somewhat tough process :) also a good article about phase:soundonsound.com/techniques/phase-demystified – frcake Apr 26 at 15:11

The formant thing came to my mind too. I've often noticed that it introduces a dusty touch when shifted down. The one I use is the build in pitch correction in Cubase.

One plugin comes to my mind: Waves DeBreath. It uses a rather special breath recognition approach. Use the output from that to sidechain duck the main vocal track. That should enhance the breathy parts.

Other than that "reverse" de-essing may work too: (upward-) Expand highs with slow energy highs.

Your Answer

 
discard

By clicking "Post Your Answer", you acknowledge that you have read our updated terms of service, privacy policy and cookie policy, and that your continued use of the website is subject to these policies.

Not the answer you're looking for? Browse other questions tagged or ask your own question.