Re: What is the best way to control espeak?

[Samuel Thibault]

Hello,

Yes, ssml is definitely the way to go, you can use espeak-ng -m -w
to directly what you want, or you can pass ssml to speech-dispatcher,
and that'll be passed on to espeak-ng.

Samuel

Jeremy Whiting, le ven. 10 juin 2022 15:08:13 -0600, a ecrit:
> Hello Rastislav,
> 
> From looking at espeak --help here I think I can answer some of this.
> 
> 1) To output to a wav file instead of speaking on audio device use the -w
> filename parameter.
> 2) To have espeak interpret SSML, use the -m argument to parse SSML (makes it
> ignore <> characters also though, so those would need to be escaped if by
> themselves in your text.)
> 3) For sound effects I'm not sure if espeak supports sound icons, but if it
> does those could be used for sound effects.
> 
> SSML can also be used for manipulating the voice parameters throughout the 
> text
> with tags that specify how the voice should change or even to change voices
> completely as far as I understand it.
> 
> BR,
> Jeremy Whiting
> 
> On Fri, Jun 10, 2022 at 2:23 PM Rastislav Kish <[1]
> rastislav.kish@protonmail.com> wrote:
> 
>     Hello list,
> 
>     this question would likely be better suited for the espeak mailing list,
>     though I'm not subscribed there and it's partially also related to
>     Speech dispatcher, so I'm posting it here.
> 
> 
>     When synthesizing text, I would like espeak-ng to do these things:
> 
>     * Change pitch in specified areas of the text
> 
>     * Change the voice variant in specified areas of the text, preferably
>     without changing the on-going intonation, though I suppose it's not an
>     absolute necessity
> 
>     * Mix in some kind of sound effect. Note that I don't mean specific
>     sounds to be played, but rather generation of various noises, brummings,
>     tones etc. in the exact time the marked text is being spoken in the audio.
> 
> 
>     It would be best if it was possible to do these things simply by some
>     markup, for example the SSML speechd modules use, though especially in
>     case of the last mentioned, that would likely be difficult to do in this
>     way, so I could do it programmatically as far as espeak could somehow
>     tell me the exact time marks an area of text is spoken, so I could
>     include the effect there.
> 
>     And of course, as far as it could give me the audio. :)
> 
> 
>     Thank you in advance!
> 
> 
>     Best regards
> 
> 
>     Rastislav
> 
> 
> 
> 
> 
> 
> References:
> 
> [1] mailto:rastislav.kish@protonmail.com

-- 
Samuel
---
Pour une évaluation indépendante, transparente et rigoureuse !
Je soutiens la Commission d'Évaluation de l'Inria.