Theiling Online    Sitemap    Conlang Mailing List HQ   

Re: Haha: Listen to Genesis in Kalaallisut

From:Henrik Theiling <theiling@...>
Date:Sunday, April 25, 2004, 13:50
Hi!

Javier BF <uaxuctum@...> writes:
> >> I used a (German) text-to-speech system programmable in Sampa to > >> compile an MP3 file of the Kalaallisut (West Greenlandic) version of > >> the Genesis text (1. Mose 1). If you want to laught, you should click > >> this: > >> > >> http://www.theiling.de/kalaallisut/bible-1mose01.mp3 > > > >More than a bit monotonous, but good. > > I loved that otherworldly echoing quality in the recording.
A problem of the synthesis. :-) Especially vowel-vowel combinations not present in German seem to be articulated by two voices for a short amount of time. That is due to the sudden switch in the middle that is not perceived as a switch but more as an echo.
> I think that, given the right predisposition,
...haha...
> listening to this recording could even induce a state of mystical > trance; in a way, it reminded me of those repetitive, solemn chants > of Tibetan monks.
Hehe! :-) I like it, too. I started to prepare the Universal Declaration of Human Rights, too. That's even longer. :-) (BTW: I found typos in the Kalaallisut version...) For some reason, one posting was dropped by my mailer. I just wanted to give a short explanation on the system used, because Mark asked. I'll make it even shorter that in the lost posting. The system was developed a few years ago at Saarland university and DFKI (German Research Center for Artificial Intelligence). I was the one to port an existing but unfinished Windows version to Unix (aweful!). The Windows versions is probably still available and is in much better shape than my Unix version, because the latter is unmaintained, messy, buggy, etc, while the Windows one was maintained for some versions. It is called Logox by GData. It can be programmed in (a variant of) Sampa, has prosody and can even sing. :-) The Unix version was used for a while in a research system, but was eventually replaced by a diphone sythesis (the original one is a microsegment synthesis). I intend to write a new version from scratch sometime in the remote future which is suitable for me and my conlanging. The old code I have is copyrighted (one reason for starting from scratch), but no-one would possibly want to look at the mess anyway (another reason for starting from scratch)... **Henrik