[MUD-Dev] DGN: Effect of voice chat on game design

[Mike Rozak]

Lost penguin wrote:

> It seems to me that a flexible TTS and/or SR engine would do a
> number of things from the immersion aspect of the game -- rather
> than having a chat window say something like "Big Nasty Troll
> says: Aaargh!", you'd actually *hear* that Big Nasty Troll, and if
> you had turned off your chat window, would then be forced to think
> "okay, is that a troll, an ogre, or something else entirely?"

While I'm all for TTS, I'll point out a flaw in your particular
example... It's difficult enough to do TTS, yet alone TTS with
enough emotion for "Aaargh!" to sound like a screem. (You can try
typing "Aargh!" into the AT&T TTS engine at
http://www.research.att.com/projects/tts/demo.html) For the
forseable future, TTS voices will be emotionless, which is ok (but
not great) for NPC conversations. Some emotion can be added using
transplanted prosody, but it's still limited. (Making TTS sound
drunk, however, is very easy.)

TTS is most useful for:

  - Text chat (which can't be prerecorded)

  - NPC conversations that can't be prerecorded (such as those
  including fill-in-the-blank elements, or automatically generated
  sentences)

  - NPC conversations that change over time (saves the servers from
  sending a copy of the wave file to every single player)

  - NPC conversations for minor characters (why bother recording a
  voice for farmer #54,435)

  - NPC conversations in general... If you mix TTS with recorded
  speech, the TTS will sound much worse than TTS by itself. It's
  like mixing live-video images with 3d renderings; the live video
  only shows how far the rendering has to go before it looks
  real. Therefore, a system which starts out with recorded speech
  will find it very difficult to add TTS, and a system with TTS will
  find it very difficult to add recorded speech. I would also point
  out that a development team using recorded speech will undergo
  much wailing and gnashing of teeth while maintaining the
  recordings database, localizing, and getting actors in a 2nd, 3rd,
  and 4th time to record new dialogues. Conversely, any VW using TTS
  will get negative comments from users about the TTS
  quality. There's no way to win, only a choice to make and
  long-lasting consequences.

Mike Rozak
http://www.mxac.com.au
_______________________________________________
MUD-Dev mailing list
MUD-Dev at kanga.nu
https://www.kanga.nu/lists/listinfo/mud-dev