[ruuk]

LOL, I hope to get quite a few people using XBMC, So we will probably find out, lol. If its anything like the Google voice hole, could last a while. But doesn't' that cause a lot of delay sending the text to external server, converting the file to wav, then having xbmc play the wav? Just some thoughts. As for the Wav's crashing, Im confused, Is it that the Wav files are to large or to small? Can we have the TTS engine generate an alternate format? or Maybe better yet, having something outside XBMC play the wav.

If you read some of the later posts after yours, it seems to be related to duration rather than size, and popcornmix has added a fix.
While it is possible on the Pi to bypass XBMC for outputting the sound, most XBMC installations currently don't have alsa enabled by default on the Pi. So for users who aren't tech savvy enough to install alsa on RaspBMC (for OpenElec, I think it has to be compiled with it enabled, and it isn't by default) we have to misuse the playSFX() command in XBMC to output sound.

I am going to go ahead and purchase a PI, So ill give the the various speech and the Cepstral option a try let you all know. One thing I don't know about the Cepstral option is will it work on the Openelec linux version. But im going to try to get a hold of them and inquire into more about it.

Cepstral needs alsa to work, so with a current normal build OpenElec it won't work. What is nice (at least in this case) about Cepstral, is it installs via a script, so it should at least install fine on OpenElec. Of course that doesn't mean it will work, and I don't really want to spend $30 to find out

Love the list count ideas, Just on of those things I didn't even realize I needed, LOL. Now that I have it, its like, Hmm, that was missing.

Yeah, I definitely need people to tell me these things. Even if I test with my eyes closed, I still have a visual memory of what the screens look like, so it isn't the same.

Ruuk, I understand your point about the xbmc team and troubles with addons and support. But I would still love to see this be as easy to install for folks who need a tts engine as it is for everyone else. My thought was maybe the team would be willing to support on some level even if just a little on the programming side of things maybe even find ways to help it along, I think what I am saying is this brings a whole new level of access for us many folks. Any time I see something as astoundingly amazing as this addon, one of my worries (And I hope not to much of my programming ignorance is showing here.) is for the future, I.E. I pray they don't close or change access to infolabels and break anything else that might drastically change the landscape. Yes I realize maybe my worry is little grandiose. But mind you there is also another side of the coin, maybe there are things they can do to help the addon as well. I mean you folks have not just created some random addon, You have actually opened up a whole new side to XBMC, From accessibility (Blind, dyslexic, reading impaired, and list goes on), to people who might just wish to run it headless for some reason or another i.e. an awesome music server. And isn't that a major ongoing goal of Team XBMC to paraphrase "to make XBMC and its user interface feel even more intuitive and user-friendly for its end-users" and "User-friendliness is next to godlyness", Well how does this not help with that goal in a seriously major way. I think what im saying is it would be cool if this were not just an addon, but would love to see it as an possible actual feature built into XBMC, You know like the 3d is now, lol. I apologize if i seem maybe a touch over passionate, lol.

Well, I'm not saying that it won't get included into XBMC ever, but I do think we are a while away from that. The addon certainly isn't ready now

My thinking had been to do as much as possible with the addon, and when we've gone as far as we can, we'll know where we actually need changes in XBMC. If we can get those sorts of changes, then perhaps later we can get things that would improve the function of the addon - for example remove the need to parse XML to speak non-focusable controls (which is slow).

I don't think you need to worry about infoLabels and such going away, because they are an integral part of all the skins. Also, one of the great things about XBMC is that it is a open community project. It's not like closed source and paid software where your issues are ignored because they don't affect the top 95% of users. If some change is coming that will break something, we can work with the developers to find a solution.

Of course, on the flip side of the coin, adding speech directly into XBMC would be a major project. One of the benefits of the AppleTV was that they could decide from the start to have speech built in. This is certainly the best and easiest way to have speech in a program. XBMC (like most open source projects) was started by a tiny group who wanted to do a specific thing, and in their case that was to get a media player on the XBox. So at the start speech wasn't even possible, let alone on the minds of the developers. Also like most open source projects, XBMC has grown as developers saw a need (generally one that was relevant to them) and filled it.
My point is that this is why adding speech directly into XBMC now would be such a time consuming task. You would need at least one developer who was committed to making it happen, and currently there is no one, and we can't expect the current developers to devote a major amount of time to it, when they are already giving their time with whatever they are currently working on.
That being said, I have no doubt the will help if they can to make what we having going here work.

As XBMC was not built from the ground up with speech, I believe that an addon is definitely the right place for TTS. There is nothing wrong with that, much of what XBMC does is handled by addons. All the skins for XBMC are addons. In fact the push over the years has been to move whatever can be handled by addons into addons.
It is a testament to the forethought of the developers that this addon has gotten this far. Despite all the things I wish I could do that I can't, there is an amazing amount of power and flexibility with the addon system, and this will only improve as time goes on.

So for now you just have me and other are willing to help. Fortunately I seem to be willing to spend ungodly amounts of time on a project that I have no personal use for Of course I write programs because I enjoy programming and learning new things. It also turns out that all the users of this project are friendly and helpful people, which really makes a big difference as well.

Anyway, I believe it fairly likely we can eventually get this addon included into XBMC, but it will take time to get to that point. I can guarantee it will take longer than you want
In the shorter term, we can definitely get this into the official repository. We just need to get it fairly complete and stable.

I was making point about XBMC and supporting addons because this addon has a large potential for having support issues. If the addon fails to speak on startup for a non tech-savvy blind user, I assume it will be as if XBMC started up with black screen for sighted user. Normal addons just have issues and throw an error or whatever when something is wrong, and don't affect the ability of the user to continue to use XBMC. So with this addon we have to do a lot to make sure that doesn't happen as much as possible, and for inclusion into XBMC even more so, but that is ultimately my goal.

[END OF LONG WINDED RAMBLE]

---

Just so I understand, was this a fix in the code for the Pi sound specifically or in code that is used on all platforms.

It is Pi specific code that is not yet in master branch. It is in the branch that gotham openelec betas build from,
and I think gotham-raspbmc-release raspbmc builds also come from there. It was in this commit (which has now been fixed)
https://github.com/popcornmix/xbmc/commi...6e78532172

The gotham release builds of raspbmc and openelec will have the fix.
You are probably best testing with the next milhouse (openelec) or miappa (raspbmc) build that appears, and they should be fixed.

It looks like openelec has just picked up the fix:
https://github.com/OpenELEC/OpenELEC.tv/...0040L15193

so it should be in next official gotham openelec build.

Thanks for taking the time to look into and fix this.