TTS would in fact be better than having the samples recorded and then compressed and sent to the game. You could maybe even do procedurally generated voices spe

TTS would in fact be better than having the samples recorded and then compressed and sent to the game. You could maybe even do procedurally generated voices speaking procedurally generated text. The quality of TTS is already better than the quality of that codec.