I use speech recognition software daily, and have been doing so for a number of years now, initially IBM’s ViaVoice (mid-90s), experimentation with other less-heard of packages, experimentation with Windows Speech-to-text when Xp appeared with the feature, and finally settled on Dragon’s Naturally Speaking 2 years ago.
I can’t say I’ve much experience of SR in games (but can say with some authority I’ve plenty of game experience – over 20 years’ worth :lol:), but one thing that strikes me, still to this day, is the amount of computing resources required for a stable, working solution.
From experience, Dragon is best (I use SR both in French and English, and I’ve been told my English ain’t so bad/accented :wink: – actually use Dragon more in English than French, but less errors in French than English… go figure! :roll: ) but it does need quite a bit of oomph CPU + RAM wise – surprisingly so for RAM, which makes muh more of a difference than CPU where accuracy is concerned. It’s reasonably accurate on a P-M 1.6 with 512 RAM DDR, much more accurate on a A64 3000+ with 1GB RAM DDR, but still not 100%, or even 95% for that matter.
My concern for a stable in-game solution, putting aside the closed nature of console hardware for a moment (but which you’d have to contend with as I expect more and more XB360/PS3 to use the feature), is what impact this feature would have on game performance and *possibly* what trade offs would have to be implemented to maintain the gameplay experience (FPS/bots/eye candy).
Moreover, I would imagine that there would be quite a vast amount of work to be done when you’re contemplating localisation – you can’t expect all of your market to be fluent in English, and if you were providing an English-only solution still, then you’d have to compensate for accents, to the risk of putting off basically any gamer who’s not US, UK, IE, AU, NZ etc. So, in that respect, perhaps voice recognition is not yet mainstream for cost reasons, rather than tech reasons.
But it would be nice to -say- play BF2 SP with bots with which you can interact through voice recognition (instead of using the Q key and selecting a standard msg with mouse), and even roll this feature into MP also (and in doing so, mitigate the amount of faffing about you have to do to get Teamspeak going and/or reduce bandwidth requirements of the BF2 built-in VoIP app). So, there certainly is bags of potential for the technology in games still…