Problem with voice activity detection about opendial HOT 4 CLOSED

I figured out a method to overcome this. It seems that the reason is when adding user input, the conversation floor will be handle to the user. I simply comment the line in the file DialogSystem. But it seems suspicious :)

public Set<String> addUserInput(SpeechData inputSpeech) {
        Assignment a = new Assignment(settings.userSpeech, inputSpeech);
        //a.addPair(settings.floor, "user");
        return addContent(a);
}

from opendial.

mm, I don't really manage to reproduce the problem, but it's probably due to some deadlock condition. Do you also have the same problem when using e.g. Nuance instead of Sphinx? Which other modules are you using (Mary TTS)?

from opendial.

Hi, I am using the Festival TTS component and I think this is the problem.
When I change it to Mary TTS everything is fine. There's probably some
problems with my Festival TTS code. Thanks!

On Thu, Nov 3, 2016 at 10:33 PM, Pierre Lison [email protected]
wrote:

mm, I don't really manage to reproduce the problem, but it's probably due
to some deadlock condition. Do you also have the same problem when using
e.g. Nuance instead of Sphinx? Which other modules are you using (Mary TTS)?

—
You are receiving this because you authored the thread.
Reply to this email directly, view it on GitHub
#19 (comment), or mute
the thread
https://github.com/notifications/unsubscribe-auth/AV8OZNCCGlLncw9OGKeA6pk5W_P6buVdks5q6fChgaJpZM4KeCHh
.

from opendial.

If you're still having problems, try to create threads to perform the speech synthesis (you can have a look at the latest version of opendial/plugins/MaryTTS.java, where we are now using Threads to avoid these kinds of deadlock).

from opendial.