Modifying/correcting TTS files

Ok - new problems. The spelling etc I've corrected did not solve all problems. Driving down the highway not close to the sea it suddenly directed me to take the ferry when I expected to be directed to prepare for the next exit. Guess it would be best to start all over with info, dictionary and lua files from the best possible english or german version. Would You attach those voice.zip's to a post pls.

ClausAnd wrote:Ok - new problems. The spelling etc I've corrected did not solve all problems. Driving down the highway not close to the sea it suddenly directed me to take the ferry when I expected to be directed to prepare for the next exit. Guess it would be best to start all over with info, dictionary and lua files from the best possible english or german version. Would You attach those voice.zip's to a post pls.

I would check the options in your program. I'm sure that you have a ferry checked as one of the road options.

Yes - ferry is enabled - but I was a few km from home. It also has a problem with right and left in some situations. Maybe a line is missing or something - I have no problem of redoing it all (its not that big a task) if I know the basis is correct.

I have problems with the torrent download so just one or two voice zips would be nice

I would uncheck the ferry unless you need it a lot.

Like asking a Swiss to uncheck mountains and tunnels....

a good English or German voice zip would be appreciated

See here:
Primo-Android-9.6.7
[color="#4B0082"]TTS (Text to Speech) installation procedures:[/color] section.
All the voice.zips there are identical to the WinCE versions except for one single line missing (removed on purpose) from info.ini---the absence of that line makes them unsuitable for all but the Android version, but all have the LUA and dictionary files you can look at--though some of the dictionary.voice files will not be as up to date as here:
TTS-Pro-dictionary.voice-files

Again - I have problems downloading from the torrent

Did you LOOK?
It's not a torrent.

Keep in mind, however, that there are differences in the LUA files based on their intended locale.
For instance, the US English files use Yards and Miles exclusively...and have a difficult time with "Kilometers per hour", etc.

ok - missed the link in the first line

downloaded the tts archive. the format of numbers and database lua files is not exactly editor (notepad) friendly. is it possible/allowed to reformat them?

If you reformat anything in a LUA file, you run the risk of corruption. They are compiled scripts and sometimes 'look' a certain way for a reason.
Sometimes not, but it's hard to tell when. Something as simple as an extra space (or lack of one) can kill them.

I'm a bit confused. Browsing thru the files I wonder witch entries are actually spoken. I see some entries in info.ini, a lot of sentences with inline variables (%s) in the dictionary file and functions, 'constants', fragments and sentences in the lua files. Where/how is the speech initiated and what steps do it pass before actually spoken?

You are getting into technical areas that nobody here may know the full answers as to when/why certain things happen. The people who wrote the program/scripts might know. {IF this occurs, AND this occurs THEN say this, BUT only during this, AND add this, BUT only if this has already happened, AND this is coming up, AND add distance before this....}
All of that has nothing to do with your issue. You want the correct WORDS spoken (the bits in quotes). When/why/how is irrelevant, and if you start changing variables and other parts of the scripts (which can refer back to other scripts that you cannot even see as they are not necessarily in the voice.zip files), you can create a domino effect and things will not work at all how they are supposed to.

I understand that lua files are to be handled with care and I guess all speech somehow passes them. I find - as far as I know - all spoken words / sentences in there. What I don't get is if/when the dictionary and/or init file is used and if/how the tags/constructs in there relate to lua files.

I don't know if this answers your question (if I understand it correctly), but......
The LUA files within the voice.zip files are the basis...and the basics.
Dictionary.voice and info.ini can take priority over these basics...depending on the situation.
In the same way that sys.txt 'over-rides' the basic settings within data.zip and system.ini.

However, if the basics of the LUA scripts (not the words in quotes...they can be literally anything you want and may or may not be correct depending if somebody already modified or translated something...and how good they were at it) are altered the whole process of what is spoken when, and where that data comes from can be broken. Everything relies on everything else.
If you were to see a whole 'family tree' of how all the script files are related, you would wonder how anything works at all.
Don't throw a wrench into that.