[Localization] e-speak and the L10n challenge of TTS

[Chris Leonard]

Does anyone have strong contacts with the e-speak developers or community?

As we are implementing text-to-speech (TTS) using e-speak, I think we
need to try to engage with the upstream to increase the L10n coverage
to meet our existing needs; however this is quite a different task
than hanging a PO file in Pootle.  As impressive as their list of
40-odd languages is, we have projects in something like 130 languages,
so there are definitely some gaps.

http://espeak.sourceforge.net/index.html

"eSpeak does text to speech synthesis for the following languages,
some better than others. Afrikaans, Albanian, Armenian, Cantonese,
Catalan, Croatian, Czech, Danish, Dutch, English, Esperanto, Estonian,
Finnish, French, Georgian, German, Greek, Hindi, Hungarian, Icelandic,
Indonesian, Italian, Kannada, Kurdish, Latvian, Lojban, Macedonian,
Malayalam, Mandarin, Norwegian, Polish, Portuguese, Romanian, Russian,
Serbian, Slovak, Spanish, Swahili, Swedish, Tamil, Turkish,
Vietnamese, Welsh. "

The process of adding languages to e-speak seems fairly complex:
http://espeak.sourceforge.net/add_language.html

In traditional FLOSS fashion, we should be thinking about how we can
try to improve our upstream.  I'm copying the L10n list to see of
there is anyone that is interested in taking on the technical
challenge of learning more about adding or improving and e-speak
language as a "guinea pig" (test subject) and sharing their experience
with the rest of the list.

This might make for an interesting collaboration between a localizer
and a developer with an interest in the importance of TTS for
accessibility in the visually impaired.  The contributions back to the
upstream of more langauge coverage will potentially have an impact far
larger than just the Sugar Community.

cjl
Sugar Labs Translation team Coordinator