Deep integration with speech synthesizer

[gabordemooij]

i.e. say: "hello"

etc.

but also some automatic stuff, reading text from all images etc.

[gabordemooij]

Can't find a decent portable synth. So now we just use an environment var and pass text to it.