2014-03-17, 16:42
It should.
2014-03-17, 17:01
I haven't tested everything yet, but XBMC seems to at least talk with Alex under OSX! This is pretty friggin' awesome!
2014-03-17, 17:48
My totally blind co-worker would love to test this out but he can't go to settings, add-ons, install from zip because he needs speech to set this up and of course that is exactly the add-on he'd be trying to install. I was able to install the add-on visually because I have some vision but many people don't and we'd be missing out a significant testing pool without those users. How would one go about installing the new addon from ruuk's repo without sight? Could we make a portable version of XBMC with the addon already installed or add the repo and TTS addon to an installer to give to testers? Thanks!
2014-03-17, 19:54
I was thinking about this today.
You can install it manually by downloading the zip and extracting it into the addons folder.
We could definitely create a windows installer that installs it. Something could be done for other platforms as well.
While it would be possible to create a bundled version, someone would have to maintain it, updating whenever xbmc or the script changed versions, not to mention that if you want to support multiple plaforms, that's a version for each platform.
Do you think having an installer would be enough?
So the user would download XBMC and install, and also download the TTS installer and install?
For the moment though, here is how you do it manually:
Download the zip:
https://ruuks-repo.googlecode.com/svn/tr...0.0.19.zip
You then extract it into the XBMC addons folder:
On Windows XP:
<root>\Documents and Settings\<username>\Application Data\XBMC\addons | example c:\Documents and Settings\ruuk\Application Data\XBMC\addons\
Vista and Above:
<root>\Users\<username>\Application Data\XBMC\addons | example c:\Users\ruuk\Application Data\XBMC\addons\
Linux:
~/.xbmc/addons/
Note that Application Data on windows is a hidden directory. You can get to it in XP by going to start::run and entering %APPDATA% into the dialog, which will open the folder.
Some extraction programs, such as winrar, will show hidden directories when choosing an extraction destination.
Then start XBMC and that should be it. You can then install the repo from the zip file to get updates automatically.
You can install it manually by downloading the zip and extracting it into the addons folder.
We could definitely create a windows installer that installs it. Something could be done for other platforms as well.
While it would be possible to create a bundled version, someone would have to maintain it, updating whenever xbmc or the script changed versions, not to mention that if you want to support multiple plaforms, that's a version for each platform.
Do you think having an installer would be enough?
So the user would download XBMC and install, and also download the TTS installer and install?
For the moment though, here is how you do it manually:
Download the zip:
https://ruuks-repo.googlecode.com/svn/tr...0.0.19.zip
You then extract it into the XBMC addons folder:
On Windows XP:
<root>\Documents and Settings\<username>\Application Data\XBMC\addons | example c:\Documents and Settings\ruuk\Application Data\XBMC\addons\
Vista and Above:
<root>\Users\<username>\Application Data\XBMC\addons | example c:\Users\ruuk\Application Data\XBMC\addons\
Linux:
~/.xbmc/addons/
Note that Application Data on windows is a hidden directory. You can get to it in XP by going to start::run and entering %APPDATA% into the dialog, which will open the folder.
Some extraction programs, such as winrar, will show hidden directories when choosing an extraction destination.
Then start XBMC and that should be it. You can then install the repo from the zip file to get updates automatically.
2014-03-17, 20:20
Obviously this is not ideal for the non-tech-savvy users.
2014-03-17, 20:34
(2014-03-17, 17:01)byron27 Wrote: I haven't tested everything yet, but XBMC seems to at least talk with Alex under OSX! This is pretty friggin' awesome!
Does it interrupt speech when you change controls? I've tested it on an OSX virtual machine, but the sound is terrible and I could only barely tell that it was speaking.
Timing and such may not be perfect yet, and different speech backends will perform better than others. We will be able to tweak and improve as we go along.
Right now I'm mostly working on getting it to speak all available text, and I think pvagner is working on ensuring that it will work with other languages at the moment.
2014-03-17, 21:34
pvagner: I added an assert in TTSService.sayText() that ensures unicode, and then made the necessary changes. Hopefully this hasn't duplicated anything you've done. Now the backends will always be passed unicode text.
Just wanted to share some other ways of communicating:
I'm on IRC at #ruuk on freenode
Saying my name in the channel will send a push message to my phone.
I can also be contacted on skype. PM me if you want my skype name.
Just wanted to share some other ways of communicating:
I'm on IRC at #ruuk on freenode
Saying my name in the channel will send a push message to my phone.
I can also be contacted on skype. PM me if you want my skype name.
2014-03-18, 03:02
So I am in Ruuk's chat room and I tested OSX and it does seem to interrupt speech when you move from one element to another. Great work! I am playing with XBMC on my windows box with NVDA and also SAPI and both work really well! If you want me to test something specific in either Windows or OSX please let me know. :-)
2014-03-18, 03:24
Looks like NVDA 2014.1 has been released. I am about to download it to see how it affects the operation of XBMC if at all. You can download it at http://www.nvaccess.org/download.
2014-03-18, 23:39
Just had a chat with amet in my channel. I was contacted by sraue, the maintainer of openelec who said he was contacted by a blind user (byron?) and that he told them he would try to do a build of XBMC with the addon included. Anyway, apparently he was having problem with no audio output from the backends in whatever setup he had, and he had concerns about platforms that don't use alsa/pulseaudio.
He wanted to know if we could output the sound via XBMC. I told him yes, but playSFX() was the only viable option, and not that viable because you can't interrupt it. He said he wanted to talk with amet to see what could be done. After a bit, he had amet join the channel to explain what was needed. In the end I asked how hard it would be to add an option to stop the playSFX() audio. A few minutes later he sent me a patch via pastebin with stopSFX() added. I'm off to build XBMC from source for the first time and try it out.
He wanted to know if we could output the sound via XBMC. I told him yes, but playSFX() was the only viable option, and not that viable because you can't interrupt it. He said he wanted to talk with amet to see what could be done. After a bit, he had amet join the channel to explain what was needed. In the end I asked how hard it would be to add an option to stop the playSFX() audio. A few minutes later he sent me a patch via pastebin with stopSFX() added. I'm off to build XBMC from source for the first time and try it out.
2014-03-19, 00:17
Good luck compiling that thing, I don't envy you! You'll have to tell us how hard it was when you are finished. 
2014-03-19, 11:54
(2014-03-18, 23:39)ruuk Wrote: He wanted to know if we could output the sound via XBMC. I told him yes, but playSFX() was the only viable option, and not that viable because you can't interrupt it. He said he wanted to talk with amet to see what could be done. After a bit, he had amet join the channel to explain what was needed. In the end I asked how hard it would be to add an option to stop the playSFX() audio. A few minutes later he sent me a patch via pastebin with stopSFX() added. I'm off to build XBMC from source for the first time and try it out.
Yeah awesome!
Still I think we will have to rework our TTSBackends to fully take advantage of this plus we might consider adding an option regarding output device selection.
For example on my HTPC I am using speech-dispatcher for TTS output and I do have dedicated soundcard with dedicated headphones for listening to that output so speech synthesis will not talk over the playing media through the main speakers.
Most likelly I'll try building this my-self on arch linux in upcoming days.
On smal devices like those powered by openelec playing through XBMC audio chain is an advantage.
2014-03-19, 14:52
I did get it compiled an it did work. I was having some issues with sluggish response in my tests, but when I woke up this morning I knew the reason, so I think it will work fine. Of course other than special builds, this won't be in an actual release for another year or so.
I was thinking of adding a setting 'Prefer XBMC Audio' or something (defaulting to true), and adding an 'xbmc_audio=False' to the available() method.
When the backend is set to auto, it would run through with it set to true and if that failed, it would set it to false and try again.
This would be in addition to what you mentioned about selecting audio devices.
I was planning on reworking the settings (after I finish doing something with stopSFX) and adding that in along with a speed setting.
I was thinking of adding a setting 'Prefer XBMC Audio' or something (defaulting to true), and adding an 'xbmc_audio=False' to the available() method.
When the backend is set to auto, it would run through with it set to true and if that failed, it would set it to false and try again.
This would be in addition to what you mentioned about selecting audio devices.
I was planning on reworking the settings (after I finish doing something with stopSFX) and adding that in along with a speed setting.
2014-03-19, 20:44
pvagner: I got interruptible speech working properly using stopSFX() from the patch. I'll attempt to submit a pull request, which will hopefully get it into the next release after Gotham. I sent a message to sraue of OpenElec about it. I assume he will be able to include it into the build he was talking about making for blind users so that it can used even sooner there.
