So I have a character who is partially a robot. Her voice is very similar to EDI in this clip: The way I describe it is, "electronically synthesized" and "heavily modulated." Not sure those 2 descriptions make sense to anyone except me though. I really do want to describe it if possible.
I read your whole post before listening to the video, and EDI's voice wasn't what I expected from your description. (Grain of salt for what follows as I'm no expert on sound engineering.) I'm unfamiliar with the universe of Mass Effect, but out-of-universe, EDI's voice is clearly a recording of a real voice, not something synthesised. I wouldn't call it heavily modulated either; in fact, I think it sounds very much like what you might expect from a human with perfect diction (ignoring opinions on accent, etc). In fact, I think they've done so many subtle tweaks with audio software to make it sounds 'perfect' (evening levels, flattening out natural spikes in volume, etc) that it now sounds 'too perfect' and has entered Uncanny Valley territory. As if whoever in-universe had made the AI's voice had prioritised clarity over realism. So I'd focus on that. If you want ideas, my description would be something like 'A voice electronically polished to such perfection that all the humanity had been drained from it'. Some other subtleties: I think the game's sound engineer has also applied one or two equalisation filters to slightly emphasise some less-natural sounding frequencies (like how your voice sounds tinny over a phone) to make EDI sound like s/he's emanating from a computer/speaker system. So filter might be a word you'd consider using in a description. There may also be a subtle delay effect (like an incredibly fast echo) applied (presumably, out-of-universe, to make EDI sound less human; I can't think of an in-universe reason). All that aside, I'll play devil's advocate and wonder whether it really matters if the reader imagines it as you do. I was perfectly happy imagining an 'electronically synthesized' and 'heavily modulated' voice until I realised that my impression didn't match yours, and (assuming your medium is written word) your readers won't have the video link like I did!
I have text-to-voice software called NaturalReader 14. I use it heavily as a consulting engineer in making structural assessment videos for clients that would rather have a little movie about their building in addition to a written report (yes, there are people out there like that). NR is 100% computer generated TTS (text to speech), and I researched the subject heavily two years ago when I was looking for good software. The references I have to speech synthesis are pretty technical, but if you want to know the terms and lingo that scientists and engineers use when describing their own efforts in synthesizing the human voice, I can point you to a website.
"A cold, clear voice with a synthesized edge. Perfect diction. No soul." or something like that. The tone depends on how negatively you want to portray the AI impression, and the character's feelings about it.
I would simply describe it as a "robotic female voice" because who hasn't seen/ heard at least one cliché robotic female in a sci-fi show that sounds just like that (or at least very similar)? I feel that, words like electronic and robotic are simple and easily identifiable by any reader.
