Share information between two rhasspy profiles

Help
1

Hello!!

My setup is:

Rapsberry Pi 3B+ running Raspbian Buster
Rhasspy 2.5.9 on Docker

I’ve two different questions.

First, let’s say I’ve rhasspy with two different profiles (language: en , hi) and I want to transfer some processed info from en to hi profile for final process or to do some action or vice-versa.

Scenario:
STT is in English and TTS is in Hindi language or vice-versa. So input would be english text or wav file and output would be in hindi TTS.

Is it possible to do?

Secondly, Is it possible to use two different TTS like Flite or Opentts or any of them listed in document in same rhasspy profile (i.e. Only en version) but one TTS would be for english and another would be for hindi TTS.

Currently, I’m using nanotts for English TTS and espeak for Hindi TTS on one rhasspy profile (i.e. Only en version) but as we know espeack doesn’t sound natural.

P.S. I’m using espeak via pip install along with pyttsx3.

So, is there any way to use Flite or Opentts same way (pip install)without any hassle of two profiles or via http where I can define which TTS when should be used.

Thanks in advance.

2

Any one, any idea?

3

If it helps, you can override the voice used for espeak by setting text_to_speech.espeak.voice in your profile JSON. When this is empty (the default), Rhasppy just uses the profile language.

The tts/say message contains a lang property that you can use (for some TTS systems) to override the voice for one sentence. Do you always want it to speak Hindi, or just sometimes?

By the way, if you happen to be a native Hindi speaker, I’d be interested in working with you to get a Hindi voice put into Larynx :slight_smile:

4

As with espeak, you can choose another voice (aka language) for Google Wavenet as well.
The response text you send should be the desired language obviously.