Text to speech in games?
by David J Weaver · in General Discussion · 02/20/2002 (6:30 am) · 10 replies
ok heres a link to the microsoft speech SDK 5.1:
http://www.microsoft.com/speech/sapi51/SpeechSDK/sdk51.asp
now is it possible to implement this into a game? would it be a good idea, or too much hastle to do?
just imagine you walk up to a friend and type hi! (in a RPG) and a little text bubble shows up atop your head and the word is then said (in a limited radious of course) i think that would be AWSOME! instread of reading all the text in a game it on-the-fly reads it to you.
http://www.microsoft.com/speech/sapi51/SpeechSDK/sdk51.asp
now is it possible to implement this into a game? would it be a good idea, or too much hastle to do?
just imagine you walk up to a friend and type hi! (in a RPG) and a little text bubble shows up atop your head and the word is then said (in a limited radious of course) i think that would be AWSOME! instread of reading all the text in a game it on-the-fly reads it to you.
About the author
#2
Sadly, TF2 may never see the light of day, so you may never get your answer from that.
*goes to harass Valve into giving him the source code*
-Evan
02/22/2002 (3:07 pm)
I think Valve was doing something like this with TeamFortress2 (wheeeereee is it?). There were certain words the game recognized, and if you typed any of them in it would say them, and I even think connect them so they sounded good (as opposed to the Steven Hawking voice).Sadly, TF2 may never see the light of day, so you may never get your answer from that.
*goes to harass Valve into giving him the source code*
-Evan
#3
I think Papyrus did something similar about 4 years ago with IndyCar Racing 2. You could communicate with the pit crew via your microphone and the system would recognise commands such as "re-fuel", "tires", and "repair". I think the feature was only available on the Mac version for some reason so I don't know how good it was. I don't know of a game that used voice recognition before it so it may have been the first one.
02/22/2002 (3:54 pm)
Looking at it the other way, I've been thinking of implementing a speech to text system using the Speech SDK where you can communicate with characters using simple commands via a microphone. But as Mike says, performance is always an issue with that sort of thing.I think Papyrus did something similar about 4 years ago with IndyCar Racing 2. You could communicate with the pit crew via your microphone and the system would recognise commands such as "re-fuel", "tires", and "repair". I think the feature was only available on the Mac version for some reason so I don't know how good it was. I don't know of a game that used voice recognition before it so it may have been the first one.
#4
02/22/2002 (4:25 pm)
Hmmm ok well my counterpoint is the fact that the req's for Text-To-speech is 150 Mhz so its no biggie. and i wouldnt use Microphones at all not like Dynamix (no offence but not everyone has got a mic. and not everyone has an Awsome one (which is what T2 requires otherwise it sucks)). just think about a RPG... if you dont have a mic you could say ":hello" and it would read it as : = speech (in a proximity) and to pronounce anything after :. so it would say hello after it printed a small text box/bouble with the same text inside of it.
#5
As was mentioned earlier, you're much better off using voice direct from the player. Most games are having that functionality, and I expect it to be within the GG community within a few months at most.
As for implementing major features, please don't come begging for stuff to be done. Only times things will get done "for you" is if the programmer was interested in it to begin with. Don't go begging for someone who isn't a part of your team to add major features for you, if you must then please give people a reason to help you out (pay, major credit, or services in return)
02/22/2002 (6:50 pm)
Unless your protagonist is Steven Hawkings text to voice sucks.As was mentioned earlier, you're much better off using voice direct from the player. Most games are having that functionality, and I expect it to be within the GG community within a few months at most.
As for implementing major features, please don't come begging for stuff to be done. Only times things will get done "for you" is if the programmer was interested in it to begin with. Don't go begging for someone who isn't a part of your team to add major features for you, if you must then please give people a reason to help you out (pay, major credit, or services in return)
#6
i learned a long time ago not to beg for other peoples stuff. im just curious if you guys think its a good idea or if i should go for something else.
02/23/2002 (5:50 am)
hehe ok Matt just to let you know im not begging i just want your oppinion on the Text to speach stuff.i learned a long time ago not to beg for other peoples stuff. im just curious if you guys think its a good idea or if i should go for something else.
#7
02/23/2002 (5:24 pm)
Valve's text-to-speech in Halflife only sounded good because the words were individually recorded by a human. They didn't quite string together smoothly, but they were fine for those Black Mesa announcements. ("Did you understand that last announcement?" the scientist asks Gordon.) I still get a thrill when I hear "You have the enemy flag" in that voice.
#8
Either way, it's more trouble than it's worth. Unless you're willing to spend an inordinate amount of time, it won't come out "realisticially". Either record it ahead of time (with a human voice) or use recording on the fly.
I know that's what we're going to be using. As for people who say "not everyone has a mic"... mics cost 5-15 dollars for the adequate ones. Mine was $5.99, and it works flawlessly. If you dont' have a mic and you want to play FPSes online you better get one! Now without the need for Roger Wilco when playing TFC it's really nice just to login and taunt other people in your own voice.
I'm just surprised why we haven't seen this in MMORPGs yet...
1. User speaks
2. Server processes it
3. Sound is played from user's location and to a set radius around them (yell = higher, whisper = lower)
This exact system is something we're going to look into for the project I'm working on. Enemies who are in the "vocal" area can listen in on your radio conversations. Sounds are already played mono, so just play them as a 3d sound witht he postion of your player. Viola! Realistic speaking in a game!
02/23/2002 (5:53 pm)
I believe the "voice" in HL was in fact created by a computer. You can find many programs out there that will convert text to sound for you, and some really dont' sound to bad (I tried one, ReadEase or something... had 3-4 voices that sounded pretty good)Either way, it's more trouble than it's worth. Unless you're willing to spend an inordinate amount of time, it won't come out "realisticially". Either record it ahead of time (with a human voice) or use recording on the fly.
I know that's what we're going to be using. As for people who say "not everyone has a mic"... mics cost 5-15 dollars for the adequate ones. Mine was $5.99, and it works flawlessly. If you dont' have a mic and you want to play FPSes online you better get one! Now without the need for Roger Wilco when playing TFC it's really nice just to login and taunt other people in your own voice.
I'm just surprised why we haven't seen this in MMORPGs yet...
1. User speaks
2. Server processes it
3. Sound is played from user's location and to a set radius around them (yell = higher, whisper = lower)
This exact system is something we're going to look into for the project I'm working on. Enemies who are in the "vocal" area can listen in on your radio conversations. Sounds are already played mono, so just play them as a 3d sound witht he postion of your player. Viola! Realistic speaking in a game!
#9
It's obviously not appropriate for a story driven, singleplayer game. But it hasn't been done before... maybe it will eventually be a staple in multiplayer?
02/25/2002 (3:18 pm)
I think this would be a good idea for an online teambased game.It's obviously not appropriate for a story driven, singleplayer game. But it hasn't been done before... maybe it will eventually be a staple in multiplayer?
Torque Owner Mike Nelson
But it seems like it'd be easier on the player (and for the game) just to record their voice then play it back to the people around him/her, rather than having them type in their messages. But this could be a cool atmosperic effect.