TTS algorithms (Re: Comments on the Text to Speech "algorithm")

[Jonathan Duddington]

On 28 Feb, marc <sintsixtus at gmail.com> wrote:

> My problem is that my own language (Dutch) does not sound very good
> in synthesised speech ..

Do you mean with eSpeak?

That's not surprising.  I've received very little feedback for the
eSpeak Dutch voice.  I am sure that there are many pronunciation errors
(eg. wrong vowel sound, stress on the wrong syllable, etc).

These problems will also affect the Dutch MBROLA diphone voices if they
use eSpeak for the text-to-phoneme translation. (information at:
http://espeak.sf.net/mbrola.html).

Compare the size of eSpeak's text-to-phoneme translation files for
Dutch (dictsource/nl_rules and dictsource/nl_list) with those for
Afrikaans (dictsource/af_rules and dictsource/af_list). I expect should
have similar complexity, but the Dutch data has a long way to go.