Lukas Loehrer wrote: > Willie Walker writes ("Re: eSpeak support in Orca -- what is the best way?"): > >> We recently looked at making a gnome-speech driver for eSpeak, but the >> main problem is that the eSpeak libraries have no facilities for sending >> samples to the audio device. Instead, it relies upon the application to >> manage the audio. >> > > I guess this is exactly why something like speech-dispatcher is a very > good idea in the long run because it spares implementers of tts > engines from dealing with the quirks of audio output architectures. > Having written code for speech output with Alsa myself, I know this is > not easy to get right, partly because low latency and instance > interruptibility of playback is important in the tts scenario. Having > this problem solved once and for all in a layer like speech-dispatcher > will give people more time to focus on actual tts code. > Not sure I agree - the model where the TTS engine delivers its sample to a "third party" for dispatch to the audio subsystem can increase latency, rather than reduce it. I do agree that there are other problems which such an architecture can avoid (for instance audio device contention).
A key requirement is the ability to quickly stop an utterance. This can be difficult with multiple processes in the mix. A corrollary to this is the need for good progress/speech marker support, since in many scenarios one needs to know how much was actually spoken before the interruption took place. Bill > Best rgards, Lukas > > _______________________________________________ > gnome-accessibility-list mailing list > gnome-accessibility-list@gnome.org > http://mail.gnome.org/mailman/listinfo/gnome-accessibility-list > _______________________________________________ gnome-accessibility-list mailing list gnome-accessibility-list@gnome.org http://mail.gnome.org/mailman/listinfo/gnome-accessibility-list