Speech To Speech

Introduction

rai s2s provides tools and components for voice interaction with the system. This package contains plug-and-play Agents which can be easily integrated with Agents provided by rai core, as well as custom ones. It also provides integration with host sound system, which can be used for low level sound manipulation.

Core Components

Component	Description
Agents	Agents in rai s2s provide functionality for voice interaction with the rest of the system.
Models	rai s2s provides a models which can be optionally installed and utilized by the Agents.
Connector	The sounddevice connector allows for interfacing directly with sound devices for asynchronous sound IO.

Best Practices

When utilizing S2S features:

1. Deployment of SpeechToSpeechAgent is meant for local setup, while the SpeechRecognition and TextToSpeech Agents are meant to be ran on separate hosts.
2. Note that sounddevice python API has notable issues in multi-threaded environment - this can lead to issues when developing Agents using the SoundDeviceConnector