TTS voices

[Menion]

Hi guys,

I'm currently in need of improved version of TTS commands. Example in worst language ever - Czech .

- 1 metr
- 1.1 metru
- 2 metry
- 5 metrů

Problem here is, that current TTS fiels do not have support for some decimal distances. And I need them for for example 1.25 kilometers value. So I've created version 3 of TTS files where all distances for kilometres and miles now do not contains value "Approximately" as before. Instead of, there are precise values like for metres, feet etc.

I've also added new parameter "%NDec meters, 0, Inf" , where for %NDec will be supplied decimal number formatted by Locus and this will be applied to all decimal numbers. Is it usable also for other languages then Czech and English? I just need to know, if decimal number are read in your language all same, like in Czech or English. Like 1.1 kilometers, also 1.5 kilometers, also 135.34 kilometers etc.

In TTS is also added one new section for time formatting. I'm preparing one new nice feature .

EDIT: btw. if someone of you improve german TTS file, I'll publish test version today . Thanks

[balloni55]

Hello menion,
possible way of speaking in german language:

1.1 kilometers > Eins Komma Eins Kilometer

1.5 kilometers > Eins Komma Fünf Kilometer   
                  or >  Eineinhalb Kilometer

1.25 kilometers > Eins Komma Zwei Fünf Kilometer     
        note: not Eins Komma Fünfundzwanzig, each decimal numer is spoken behind the other

[Menion]

fine, in all cases, spoken distance ends with "Kilometer", fine. Same in Czech and also in English. So it should be ok then. Thanks

[ta-ka]

Hi menion,
In Japanese, it's always "kilometer" for any decimal numbers.

In TTS is also added one new section for time formatting. I'm preparing one new nice feature .

I guess this is for estimated amount of time to destination and not for estimated time of arrival. I would like to be sure.

[Menion]

Hi menion,
In Japanese, it's always "kilometer" for any decimal numbers.

Perfect, thanks. So solution should work in this case.

And time - is purpose important? Because non of your options is correct. Anyway in future it should be used for various purpose

[ta-ka]

And time - is purpose important? Because non of your options is correct. Anyway in future it should be used for various purpose

My guess was wrong...   Never mind.
I asked the question because there are different spoken words for 'duration time' and 'exact time'.
So, the various purpose contains only one type of them, it shouldn't be a problem.
But if both types are mixed, it will be a problem.

[Menion]

hmm I found same problem in Czech now. Damn. How easy English is ... . Thanks, I'll have to make TTS file even more complicated. So stay a day, two with translating, there will be some changes

[Menion]

Fine, so en.tts and cs.tts are now converted to version 3. More information about what has changed is in README file, so please read it!

Feel free to ask if something won't be clear. I'll then add more comments on required places. Thanks!

[gynta]

en.tts and cs.tts are now converted to version 3

de.tts also...

[ta-ka]

updated ja.tts to version 3.
fixed missing commas in en.tts.

[Menion]

@gynta, thanks!

@ta-ka, I'm surprised it worked for me even with these issues, thanks!

I'm for @Christian and @0709 preparing one surprise (hope it will work, do not know yet ), so test version will have a delay few days maybe. Be prepared .

[ta-ka]

If there is no need to replace the digits in "numbers_N1", is it allowed to make it empty?

Code Select Expand

},
"numbers_N1": { },
"numbers_N2": {
Or, is it allowed to completely remove it?

Code Select Expand

},
"numbers_N2": {

In ja.tts, "numbers_N1" doesn't contain the words to replace but only digits like follows.

Code Select Expand

"numbers_N1": {
      "1": "1",
      "2": "2",
      "3": "3",

[Menion]

that's what I wanted to say you. I think there have to be at least empty N1 category. Locus tries to find required "translation" (for example for "1" in English - "one") and if there is no such translation, number itself is send to TTS engine. In languages where numbers are always same it is not a problem.

It is useful for languages like Czech, where exists many various of word "one" - jeden, jedna, jedno, etc.

[ta-ka]

In case of popular freely available Japanese TTS system in android like Google TTS or KDDI N2TTS, passing numbers as digits to the TTS system gives satisfiable results for cardinal numbers, as the TTS system takes care of different pronunciations. Only ordinal numbers have to be replaced in Locus. So, I'm going to test and update ja.tts.

It is useful for languages like Czech, where exists many various of word "one" - jeden, jedna, jedno, etc.

Knowing about other language is always interesting.

[ta-ka]

I tested ja.tts with empty "numbers_N1" as shown the first code in post #86, and got an error message below just after locus start to speak something. There was no force close. Locus kept working but voices of Navigation/Guidance/Training were complicated mixture of Japanese and English.