2014-05-04, 20:59
(2014-05-03, 21:35)jhall Wrote: I put both the 0.0.48 and speech.server.zip in the addons dir, but I'm guessing I don't need the speech.server. It doesn't look like an addon anyway, now that I look through it a bit better.Correct. The speech.server is so you can have the speech generated on another machine and either send the wav to the client or speak the text on the server.
This is especially useful for lower performance platforms like the Raspberry Pi because you can offload the speech processing to a more powerful machine.
(2014-05-03, 21:35)jhall Wrote: I must say, I'm not a fan of the keyboard input method with the ridiculous remote I got, who ever heard of a remote that doesn't have numbers on it anyway! and in the end I just went in and edited the .xml settings file and added the proper values since the ip address input method was really annoying. There did not seem to be a way to make the pi read the input field so I could tell exactly what was in there. I can't imagine the fun I would have if I had to type in a WPA key or something.This must be an issue with the Pi. I'll have to check it out. The normal text input dialog speaks the letters you type or delete, and the IP address dialog speaks the current part of the address you are entering so if you are typing 127, it will first say one then twelve and then one hundred twenty seven.
None of this behavior is set in stone, this is just how I've done it so far. It sounds like it's not even doing that for you.
(2014-05-03, 21:35)jhall Wrote: The initial dialogue boxes (welcome to openelec etc) did not properly speak, and since this remote has no f2 or f3 buttons on it I didn't try to make it read, simply clicked on next until the main gui showed up. (previously I had used google goggles to get an idea on what might be going on)There is a lot still to do getting everything to speak. I'll have too look at what that welcome dialog actually is and see what I need to do about it.
I also plan on making the buttons/keys configurable.
(2014-05-03, 21:35)jhall Wrote: Then I enabled live tv and enabled a pvr backend. There are 1,434 channels in my backend with 2 weeks of EPG for each, undoubtedly that caused the pi to be unhappy over memory. I scrolled down the list until I found a channel and clicked on it. The channel began playing, but suddenly speech was aborted and would not play I'm guessing because media was playing? Is there a way to pause the media or something so that it will speak long enough to figure out how to get out of it?Speech should work even when video is playing, so the cause of it stopping must be something else.
Unfortunately, I don't have any PVR hardware or even cable or satellite to make one useful, so it will be hard for me to duplicate the issue. Maybe I can fake something up somehow.
(2014-05-03, 21:35)jhall Wrote: My wife had walked in while I was battling with the IP address input madness and offered to help. At some point the thing stopped speaking, which was just fine for us since it was being a pest. We kept clicking on ok until we got to the 'enable' stage of turning on the addon. Since clicking enable did not work, I walked to a computer, looked in the settings.xml file and rebooted the pi.That's one option I have considered. I would have liked to have been able to have F2 speak the text, but unfortunately the keyboard dialog captures all keyboard input. I think I will implement your idea.
Maybe one solution to not reading the input is if you pause without moving the selector around for more than 3 seconds after the last sniffy sound, it will read the input value. At least then you could figure out what was going on. and then of course if you hit backspace, it could read immediately. I don't know whether the input field can be polled or if you just have to keep track of it on event or something, but it makes that sniffy noise all the time, I'm guessing you could hook onto that callback to determine what was clicked.
(2014-05-04, 13:48)jhall Wrote: I wouldn't release it just yet. Admittedly I jumped to Helix build on accident because I wanted to be sure to get all the latest fixes, and xvdr plugin is not available for either Gotham or Helix in Windows, but I did notice that under windows when you start xbmc, for some reason it routes all the OSD and programming through the headphones and the jaws through the laptop speakers. That approach in theory makes sense, but is a little annoying in a quiet environment where people are trying to sleep.XBMC sound output and JAWS sound output are unrelated to each other. The addon just tells JAWS to speak the text and JAWS plays it wherever it's configured to play it.
(2014-05-04, 13:48)jhall Wrote: Also I had to do a bit of research to get vnsi5 plugin for vdr going because a decision was made to remove the vdr side from the former vnsi3 git tree. It did compile for the version of vdr I am running ,s ono real headaches there.Within the next few weeks, I will create a release thread for the addon itself, and we can move discussion about the addon itself there and people can continue to discuss other issues here, or we could have a general blind/partial-sight support thread. I'd like to think the title of this thread is becoming somewhat less apt than when the thread was started
It does seem to take several seconds to change channels on windows, not sure why--I mean 5 or more. In contrast, vdr can change the same channels in under 2 seconds.
In windows, while playing, I can press select button to cause the OSD to appear. arrowing to the right eventually brings me to an entry that is called the channel name (the one came from the vdr server) pressing select on that brings up the EPG. Cycling through the events using up and down (choosing events from nearby channels) you can press select and the event description shows up. I used jaws OCR recognission to get a fuzzy read on the description, and then arrowing right and enter to switch caused the channel to go away for 5 or more seconds, then resume the former channel not the new one that I had selected via movie. I'm thinking that one might be an XBMC bug.
Since this is my first real exposure to XBMC, I haven't really categorized my library yet, and all my vdr videos don't show up in the library. My movies are not aptly named either, so not sure what to do about that yet.
It's not a directly-related question to the plugin development, but does raise usability questions from a blind-person perspective for XBMC. Maybe we should start a separate thread for those kinds of issues? I personally wouldn't mind them thrown in here, but I don't want to presume.