From: Jonas H <rasher_at_rasher.dk>
Date: Wed, 22 Nov 2006 08:29:11 +0100

Simon Eigeldinger wrote:
> Hi all,
>
> Are we able to create voice files automatically?
> I think it will be interesting, if we can create also daily voice
> files. So we are all up to date.

With my somewhat recent voice update, this should be at least
technically possible. The voice building I added supports a number of
different TTS engines: Festival, Flite and espeak, so this should
certainly be feasible. Rather than daily, it should be enough to rebuild
every time the language files are updated.

> I found quite nice synthesizer engine which can run under linux:
> http://sf.net/projects/espeak/
>
> I don't know how much processor power we have available on the
> players but maybe we can port this lightweight speech engine to
> rockbox? If this is possible we don't need voice files any more.
> The speech engine can also then read text files on the player.

I believe Marcoen Hirschberg (markun) is already working with the eSpeak
author on doing exactly that. It's extremely unlikely to work on the
Archos units, and might not be feasible on the current targets either,
but the in-progress Toshiba Gigabeat would probably be perfectly realistic.

-- 
Jonas H
rasher(at)rasher(dot)dk

Received on 2006-11-22