[mtsonline]

Hi together,

absolutely awesome, thanks! I followed the steps to install alsa and activated it. Installed espeak and the newest xbmc-tts version from the repo.
And the voice came up emedeately, no problems so far. :-)
I will keep on testing it. Currently I am using rasbmc on the pi.
When I have some free time I would like to test the google tts. Would I just have to install the https server from the repo and activate that or how is this done?

Thanks again for this genious work! Maybe you could contact the rasbmc developer and ask him to include the addon and dependencies right out of the box, this would help a lot of people get access to a new world of multimedia.

looking forward to further improvements and development,
Martin

[ruuk]

When I have some free time I would like to test the google tts. Would I just have to install the https server from the repo and activate that or how is this done?

Install either mpg123, mpg321, sox or mplayer and it should show up as an available engine in the addon settings. I'm not sure which of these are available on raspbmc off the top of my head.
The http server is separate backend.

[Traker1001]

Thanks again for this genious work! Maybe you could contact the rasbmc developer and ask him to include the addon and dependencies right out of the box, this would help a lot of people get access to a new world of multimedia.

looking forward to further improvements and development,
Martin

When I have some free time I would like to test the google tts. Would I just have to install the https server from the repo and activate that or how is this done?

I forgot to add the step to get google tts on Raspbmc,
Type "Sudo apt-get install mpg123" in the shell, Without the qoutes.
Restart XBMC and GoogleTTS should be an choice for one of your speech engines.

I actually converted to RaspBmc myself. So I can't yet tell how to update the dependices on Openelec PI, They lock the platform down with Openelec Pi, I may look into creating a custom build myself since they do provide all the tools. If I do so successfully, I will make sure it is available to everyone.

I don't see the developers of RaspBmc adding alsa by default yet, as being it is highly experimental still. (I could be wrong though)

I just got the same behavior. It happened when I switched skins. Reading the categories is something that is specific to quartz. The addon only detects the skin when it first starts up, so when I changed to Quartz, it still thought I was on Confluence. Could that be what happened to you?

I just updated my OpenElec on Pi to the 0.0.54 and they played for me there. Could you double check to make sure yours got updated. The only other thing I can think of is having nav sounds turned off or that something was playing and nav sounds are disabled during playback. If none of that is the problem then maybe a log would help.

Yeah, If I weren't about to crash at the time I would have realize this myself, lol. and never posted, I can verify it works

Nav sound on, But for some reason not playing on either my windows or Pi. And I mean ANY nav sounds. May be time to wipe out and start fresh again. I did bunch of messing with the menu xml files the other night and also adding stuff trying to get MBrola voices and I may have broken it. Don't believe the sound issue is plugin related. Ill get back to you on this.

RockerC, Any thoughts on also adding basic voice control command support to your addon or as separate addons to add further accessibility functions into XBMC?

While I appreciate the thoughts, There are already plugins to add voice command to XBMC, For example, I understand that VoxCommand with the Amulet remote works very well. Personally and I can't speak for the others, I have no interest in voice command, and especially it being wrapped into this plugin. There is already enough to tackle with the addon as it is without adding to a whole other level of complexity and issues both hardware related and sotfware. Furthermore, I personally wouldn't want anyone misunderstanding the purposes of this addon, And that isn't to add ease so much to XBMC, But to provide a platform for those who for whatever reason don't have access to the text components of XBMC.

[Traker1001]

A thought regarding Dependencies, I don't believe you should add any of the dependencies that can be easily installed. However I do think we should start a running list for the Wiki.

So far I have on the Pi/Linux, Espeak (Or some other voice engine, See Below), Alsa, Mpg123, Comtypes (with is automatically added with the addon now),
For windows, SAPI, Comtypes (with is automatically added with the addon now).

Also for speech engines this works with or (Back ends)
Windows: SAPI, NVDA, Speech Server (Which requires A running server hosting speech engine with voices), Jaws,
Linux: Google TTS (Needs MPG123 and ALSA), Espeak (Doesn't need Alsa, But works better with it), Festival, Flite, (Voices Suck with these 2), HTML Speech Server (See Above), Cepstral (Very Slow on the Pi only), Pico2Wav
Other Platforms: OSX?

No Speech for Android OS (Yet!)
No speech for Apple I-devices.

Let me know if I missed anything...

[ruuk]

Nav sound on, But for some reason not playing on either my windows or Pi. And I mean ANY nav sounds. May be time to wipe out and start fresh again. I did bunch of messing with the menu xml files the other night and also adding stuff trying to get MBrola voices and I may have broken it. Don't believe the sound issue is plugin related. Ill get back to you on this.

Did this start when you enabled alsa perhaps?

[ruuk]

Just wanted to point out that Cepstral is only slow (well noticeably) on the pi.

On a related note I got this email from Cepstral:

----------
Rick,

Checkin in on this since I haven’t heard back from you. Were you able to get the Callie voice working with the XBMC on the raspberry pi?

Patrick Dexter
Director of Business Development
Cepstral LLC
-----------

Just thought it was interesting.

[Traker1001]

Did this start when you enabled alsa perhaps?

No my problem is on both the pi and my windows machine, there is something in common and i have to figure it out. Whatever it is there are no errors being thrown.

Just wanted to point out that Cepstral is only slow (well noticeably) on the pi.

Ok, that makes sense. Forgot about Linux in general. to many things in my head at once, lol. Updated.

That is interesting.

[mtsonline]

Hi again,

thanks, installing mpg123 worked without any issues and google voice is great.

@ruuk
Could you add a voice change setting to the google tts? Currently I hardcoded my voice into the tts url in the google.pi. But it would be nice to have a setting for this I guess.
As I read google should have a list of available voices that can be requested through the api and the language code just has to be put instead of tl=en to be activated.

Just as an idea, I also do not worry about changing it in the code to fit my needs.


cheers
Martin

[Traker1001]

Actually, A thought I had was that maybe isolate Google Translate and/or the HTTP portion from the TTS addon for size, memory and menu complexity. Then make the TTS addon a dependency of google translate addon, but not the other way around, obviously.

Then we take google translate, We allow for the creation of a cache mp3's for the common items, I.E. Typical nav System, Addons, settings, Videos, so that those items don't need a request. Which is much of the menu system and only send the actual request to G.T. with variable menu items. This would be alot of work, however, with a tutorial and little help, maybe users would be willing to create there own cache packages and share them. This would skip alot of the need for network traffic.

It would almost be like creating our own voice package for xbmc, lol.

Of Course then again I may just and probably am over thinking it, lol.

[Traker1001]

Ok, First I don't have a mac anymore. However, I do know there are some pretty nice voices on IOS.
I came across this site that looked interesting. Wonder if we could do some slight re-purposing of this for our own purposes.

Take a look.

http://wolfpaulus.com/jounal/mac/ttsserver/

[RockerC]

Any thoughts on also adding basic voice control command support to your addon or as separate addons to add further accessibility functions into XBMC?

While I appreciate the thoughts, There are already plugins to add voice command to XBMC, For example, I understand that VoxCommand with the Amulet remote works very well. Personally and I can't speak for the others, I have no interest in voice command, and especially it being wrapped into this plugin. There is already enough to tackle with the addon as it is without adding to a whole other level of complexity and issues both hardware related and sotfware. Furthermore, I personally wouldn't want anyone misunderstanding the purposes of this addon, And that isn't to add ease so much to XBMC, But to provide a platform for those who for whatever reason don't have access to the text components of XBMC.

Actually there are no Python-based addons/plugins that add voice command to XBMC, only separate voice command applications that can send commands to XBMC, none of which are cross-platform, and no one that is close to being as simple to install as an addon, and no one that is even particularly simple to use with XBMC.

Thus I still feel that this on-topic here since this thread topic is "XBMC not working for blind users" and the discussion subject is about accessibility for the blind, which does not just mean Text-To-Speech.

I think having a (if so separate) simple to install Python-based addon/plugin for XBMC is a valid subject to discuss here as an idea generally improve XBMC accessibility for the blind.


Another proof-of-concept is Amazon Fire TV which have a very good "Voice Search" feature which shows how voice command adds accessibility to the search function in a media player

[Traker1001]

The reason being is it is one thing to send some infolabels to a speech engine and come back with a wav or mp3 then play it. Speech to command is a WHOLE ENTIRE other level of complexity, And I don't believe voice control will add anything to accessible and in fact will make it less accessible, because there is no easy way to just set it up without controlling ALL the factors, I,E, Hardware, Platform, OS, Drivers, etc.. There are far more variables in voice control that there are in text to speech.

Im sorry, I just don't believe that Voice control has a place here. However, I just do the testing, Maybe Ruuk or one of the actual developers wish to chime in.

Also, I think the thread topic itself is off topic, Text to speech isn't limited to just the blind, It branches out to disabled of all sorts. And even just those who may like to run a system w/o a display, I.E. stereo control system, music server, even home automation setup. If I could change the topic I would just have it as Text to Speech for XBMC and leave out the part for the blind.

[ruuk]

Could you add a voice change setting to the google tts? Currently I hardcoded my voice into the tts url in the google.pi. But it would be nice to have a setting for this I guess.
As I read google should have a list of available voices that can be requested through the api and the language code just has to be put instead of tl=en to be activated.

Sure, I can add this.
As far as I can tell, this just allows changing to a language specific voice, rather than different voices for the same language, correct? I just want to be sure I'm not missing something.

[ruuk]

Actually, A thought I had was that maybe isolate Google Translate and/or the HTTP portion from the TTS addon for size, memory and menu complexity. Then make the TTS addon a dependency of google translate addon, but not the other way around, obviously.

Then we take google translate, We allow for the creation of a cache mp3's for the common items, I.E. Typical nav System, Addons, settings, Videos, so that those items don't need a request. Which is much of the menu system and only send the actual request to G.T. with variable menu items. This would be alot of work, however, with a tutorial and little help, maybe users would be willing to create there own cache packages and share them. This would skip alot of the need for network traffic.

It would almost be like creating our own voice package for xbmc, lol.

Of Course then again I may just and probably am over thinking it, lol.

That's an interesting idea, and certainly helps relieve abuse of the Google server and speed up speaking for common controls. I think it's also a lower priority item at the moment, but I added it to my todo.txt file that I've added to the addon where I'm adding all these ideas so I don't forget them.

Having Google included in the addon doesn't appreciably increase memory usage of the Addon. It's only 60 lines of code. Not to mention there are ways I can make the addons not take any memory until they are in use.
Also, one of the benefits of having included is that it is an engine that may be available when others are not. Even with no speech programs installed on a computer, if any of the supported players is installed, the Google engine will be available. One of my goals has been to have this speak "out of the box" as much as possible, and part of that strategy (especially for the Unixes) has been to support as many engines as possible so that it can use whatever is installed.

[ruuk]

Ok, First I don't have a mac anymore. However, I do know there are some pretty nice voices on IOS.
I came across this site that looked interesting. Wonder if we could do some slight re-purposing of this for our own purposes.

Take a look.

http://wolfpaulus.com/jounal/mac/ttsserver/

Unfortunately, they don't include the python source, only the bytecode files.
It looks like this just uses the OSX internal speech engine, and if I just add support for changing the voice on the OSX backend, then our server would work similarly.