From: Rick B. <rb...@ca...> - 2006-01-03 05:44:24
|
Hello all, Cepstral got this working at one time under Linux, and then broke it with an update. They say they might get around to fixing it in the future, but I'm tired of waiting. Basically, the Cepstral TTS engine isn't alsa compliant, so it doesn't concurrently multiplex with other audio streams. Anything "spoken" through swift backs up behind whatever app is currently using the audio device. If I'm streaming an Iradio feed or playing an mp3 playlist, many (speech event) processes can backlog on the machine and exhaust memory. At the command-line I have a workaround: mplayer | swift "whatever needs to be spoken" ... basically piping swift output to mplayer, which is alsa compliant, and multiplex friendly. Does anyone know how best to configure the MH ini to use the "mplayer | swift..." command combo for my speech engine? Thanks, Rick |