How to make voices sound robotic?

[Morph Bark]

I'm currently working on a video game with several robot characters. We're a team of 11 students doing this as part of our education. We've got auditions running (here (http://www.behindthevoiceactors.com/members/TeddyRight/casting-call/Project-RIAS/)), but we wish to alter the voice files ourselves to ensure that they all sound robotic in the same way, not too much, and not too little. However, as few of us have experience working with audio a lot, we still need to find ways to make the voices sound appropriately robotic enough for the characters in the game. Does anyone know good software for that, or techniques that help with this?

[Grinner]

I'd go with audio filters or speech synthesis software.

The first requires just processing whatever lines you record. Google tells me that in Audacity, you'd want to use the Vocoder effect.

The second would obviate the need for auditioning, and really, your stated purpose is ideal for text-to-speech technology. Current speech synthesizers have difficulty generating anything other than robotic sounding voices. Depending on your game software, you may even be able to incorporate it into the program.

[Oneris]

For a GlaDOS-style voice, I see many people using Melodyne for pitch modulation. It's not free, but there's supposed to be a demo version that's as good as the full version, or so I've heard.

[Eldan]

Actually, the thing that comes to my mind? Train station announcements. You perhaps know the kind.

"The train to ... Zurich ... and ... Berne... will leave at .. 5... o'clock and ... fifty ... five ... minutes. There will be a delay of... six... minutes."

So, how about recording the dialogue in pieces? Just speak a few words at a time, then wait a minute, record the next part of the sentence, then string it together in an editing program, leaving slightly awkward pauses. That way, the pitch, speed and intonation will always be slightly off from word to word.

[factotum]

Over-use of autotune (as in the thing they use to help out singers who can't quite hit the right note) produces a very odd effect (you can hear it very clearly in Cher's "Believe"). Maybe use that with the note required set quite low?

[Brother Oni]

Maybe use text to speech, then plug that through different filters to get different voices?

[Jeff the Green]

Actually, the thing that comes to my mind? Train station announcements. You perhaps know the kind.

"The train to ... Zurich ... and ... Berne... will leave at .. 5... o'clock and ... fifty ... five ... minutes. There will be a delay of... six... minutes."

So, how about recording the dialogue in pieces? Just speak a few words at a time, then wait a minute, record the next part of the sentence, then string it together in an editing program, leaving slightly awkward pauses. That way, the pitch, speed and intonation will always be slightly off from word to word.

You could also combine that with multiple voice actors. Have them each read the line and alternate words.

[Gnoman]

(you can hear it very clearly in Cher's "Believe").

A bit off topic, but that was completely intentional. Autotune was initially used for special effects.

[factotum]

A bit off topic, but that was completely intentional.

I know it was--I wasn't suggesting for a second that Cher can't sing! It was just a good example of what it sounds like.

[Gnoman]

You have no idea how many times I've heard people suggest exactly that.

[Amidus Drexel]

As far as specific types of voice effects go, ring modulators are pretty common for robotic voices (think Daleks), as are... I want to say phasers?

[Ravens_cry]

A very simple, low tech way to get a tinny sound, at least, would be to record the voices, play them back and record the recording.