From: Paul W. <pa...@wi...> - 2002-09-13 06:46:06
|
Here is another way to integrate Mr House and Natural Voices on Linux. In mh.ini set: voice_text=attunix voice_text_attunix=/usr/local/naturalvoices/ATTNaturalVoices/TTS1.4/Desktop/bin/TTSStandalonePlayerDT -v0 -data /usr/local/naturalvoices/ATTNaturalVoices/TTS1.4/Desktop/data Change the path to match where you have installed Natural Voices. The at line 115 in lib/Voice_Text.pm add: $speak_pgm = $main::config_parms{voice_text_attunix} if $speak_engine eq 'attunix'; and at line 222 add: elsif ($speak_engine =~/attunix/i) { print "About to send data to natural voices - $parms{text}"; my $pid=fork; my $speaktext=$parms{text}; if ($pid == 0) { open(VOICE,"|$speak_pgm"); print VOICE $speaktext; close(VOICE); exit 0; } } This resolves the issue of Mr House pausing as the voice output is performed in a second thread, however it can take Natural Voices a while to start up, so I am thinking of creating a simple voice server that will listen on a network socket. Has any one had any success with the JSAPI examples included with Natural Voices? They seem to have a bug where no synthesizers can be found. Regards, Paul Wilkinson |