Efficient speech and channel effects using a modified low bit rate speech codec

(Or, the architecture behind TinySpeech for Games™)

The Problem

Should all aliens or other non human creatures speak perfect English? Or Japanese, or any other modern language? Good games, just like good movies, are about immersion. They can make the player forget where they are, and convince him he is IN the game, not just looking at it. In my opening fictitious example this immersion was broken by an alien message that was obviously from another human player.

Sure, some players will play along and attempt to alter their normal speech to match their character, but will everyone know what a Sorz sounds like? Or a 40' tall giant? Can a female player with a high pitched voice lower the pitch of her voice enough to make a convincing giant? Or a male player with a low pitched voice raise his pitch enough for an elf?

Do radios always work perfectly? Should a radio transmission from a distant ship sound as good as the base station that is near to you? Should your enemy attempt to jam your transmissions? Should certain types of weapons interfere with communications? Should your location relative to power sources, radiation sources, or metal buildings interfere with communications?

All of these effects, from morphing speech to adding static to a communication channel can all be done with conventional signal processing, but they require extra CPU cycles and can unintentionally degrade the intelligibility of the speech. And when they are done on the receiving end there is a possibility that not everyone will hear the same thing. The game could be hacked to remove the effects, which could give a player the advantage of listening to clear unmodified speech from all other players.

Multi-player games also face the bandwidth challenge, especially with most players still using modem connections. And companies that host massively multi-player games pay network charges based on their total monthly bandwidth usage, so adding speech chat can add greatly to that cost. This is why games use speech compression to greatly compress the speech. But speech compression also degrades the intelligibility, and adding more processing for other effects can make it TOO hard to understand. What is needed is a combined speech compression with effects so that few, if any, additional CPU cycles are needed. Those effects should be made at the sender's end to reduce processing and prevent hacking to bypass the effects on the receivers end.

Next: The Solution


Prologue and Introduction
Background
The Problem
The Solution
The Solution continued and Conclusion