Resource friendly sayit Adapter

ioBroker / AdapterRequests

This Place is used to track the status of new Adapter-Requests.

248 stars 36 forks source link

Resource friendly sayit Adapter #94

Open Marty56 opened 5 years ago

Marty56 commented 5 years ago

For each speaker class the user has to create an instance. I my case I have three instances. On small system this unfortunately comsumes a lot of memory.

there should be only one instance for the universal sayit adapter. In the configuration user can add various speaker from different classes..
This new sayit adapter should also contain support for amazon echo for consistent usage.
preferrable this adapter should support a "cloudless" text to speech conversion

GermanBluefox commented 5 years ago

Do you know some offline libs for text 2 speech?

сб, 14 июл. 2018 г., 8:06 Marty56 notifications@github.com:

For each speaker class the user has to create an instance. I my case I have three instances. On small system this unfortunately comsumes a lot of memory.

there should be only one instance for the universal sayit adapter. In the configuration user can add various speaker from different classes..

This new sayit adapter should also contain support for amazon echo for consistent usage.

preferrable this adapter should support a "cloudless" text to speech conversion

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/ioBroker/AdapterRequests/issues/94, or mute the thread https://github.com/notifications/unsubscribe-auth/AEXqgOpF9LkJ8XIC4kb-Ta8mGXyU88K4ks5uGYpOgaJpZM4VPvkG .

Marty56 commented 5 years ago

Unfortunately no. Sorry

GermanBluefox commented 5 years ago

So it will be hard with "preferrable this adapter should support a "cloudless" text to speech conversion" :)

GermanBluefox commented 5 years ago

The problem in single instance adapter is the configuration dialog and not engine itself.

Marty56 commented 5 years ago

"The problem in single instance adapter is the configuration dialog and not engine itself." This sounds easy to realize :-) But I am not a developer.

I already guessed that the "cloudless" text to speech is difficult looking at the discussions in the forum. Especially if it should work for all platforms. In the forum they have discussed a solution which would run on raspberry.

GermanBluefox commented 5 years ago

In the forum they have discussed a solution which would run on raspberry.

Link?

dwm66 commented 5 years ago

Regarding the offline TTS: I played some years ago with something called "MBROLA": http://tcts.fpms.ac.be/synthesis/mbrola.html Worked quite ok, but I remember it was not really ressource saving :) A list of TTS systems can be found on wikipedia also: https://de.wikipedia.org/wiki/Sprachsynthese