speechd-discuss
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: speech-dispatcher and piper


From: Samuel Thibault
Subject: Re: speech-dispatcher and piper
Date: Sat, 23 Mar 2024 23:32:04 +0100
User-agent: NeoMutt/20170609 (1.8.3)

Hello,

Hudson Kendall, le sam. 23 mars 2024 11:59:11 -0600, a ecrit:
> Mainly, I'd like to disable the tendency for speech-dispatcher to feed the 
> text
> one sentence at a time to piper. piper is already very good at synthesizing 
> one
> sentence ahead while it speaks the last,

So it does starts sending the beginning of the audio to the pipe before
synthesizing the rest? If so, we can indeed drop the delimiter and raise
max chunk length for this module.

> and when speech-dispatcher stops and restarts piper for every
> sentence, it causes long gaps in playback between sentences.

Long gaps are a bad sign, because it means that using it with e.g. orca
will be laggy. Is it not possible to also avoid them some way?

> I tried modifying the ModuleDelimeters and ModuleMaxChunkLength options, but
> they seemed to have no effect.

The documentation was wrong, they are actually named GenericDelimeters
and GenericMaxChunkLength.

Samuel



reply via email to

[Prev in Thread] Current Thread [Next in Thread]