Open Voice OS
What is Open Voice OS
⚫ Open Voice OS or OVOS as we like to call it in short
aims to be a full operating voice assistant stack that is
free and open source, the stack can be better defined
as:
⚫ Stack as a Framework
⚫ Stack as an Operating System
⚫ Before diving deeper into the above.. let’s see what its
built upon.
How did OVOS start ?
⚫ OpenVoiceOS originally started as heavily patched
Mycroft Stack running on an embedded buildroot
system for DIY Raspberry Pi smartspeakers
⚫ Mycroft community developers interested in
supporting more platforms and expanding the stack
⚫ Not under the control of any one organization
⚫ Community First!
The internals
⚫ OVOS consist of python based core framework called
ovos-core, this core framework is responsible for
providing distinct services
⚫ Some important services it provides are:
− Websockets Messagebus Service
− Skills Service
− Audio Service
− Speech Service
− GUI Service
The internals – Websocket Messagebus
⚫ The websocket messagebus service is our master
communication layer, everything in the OVOS world
talks to each other on this layer
⚫ Service A
⚫ Service B
The internals – Skills Service
⚫ Mini voice interface apps we call skills live along side
the intent parsers here
⚫ Provides a manager who’s sole responsibility is
lifecycle management of skills
⚫ intent engines parse text to actions here
Skills Service
Intent Manager Text To Speech
Speech To Text
Got Text - Skill A
Magic Here Parser Magic Here
- Skill B
The internals – Speech Service
⚫ Housing for the Wake Word Engine API
⚫ Housing for the Speech To Text Engine API
⚫ Responsible for everything in the speech input process
Speech Service
Record Processed Text
Wake
“Hey Mycroft” User Transcribe
Up! To Skills Service
Audio
WW Engine Always On STT Engine
The internals – Audio Service
⚫ Housing for the Text To Speech Engine API
⚫ Housing for the Playback Services / Multimedia
Framework
⚫ Responsible for everything in the audio output process
Audio Service
“Tell me Skill B Chuck Convert
Speech Skills Norris TTS
a joke” Found A CAN To
Service Service divide Output
Match by 0. Audio
The internals – Gui Service
⚫ Talks to Mycroft Gui in the gui protocol language
⚫ Responsible for skills display lifecycle management
and skills display data management
GUI Service
Skill B Chuck
Norris
Found A CAN
Match
divide
by 0.
Screen
Page.qml Page Data
Audio
Service
The externals
⚫ Apart from the core framework there other major
external frameworks that build up our stack
⚫ OVOS Shell
⚫ OCP Multimedia Framework
OVOS Shell
⚫ Kirigami based application running on EGLFS using the
Mycroft GUI protocol view
⚫ A GUI Interface developed for our smartspeaker and
embedded platforms
⚫ Uses a custom Kirigami Platform Theme plugin to
provide global themeing to all our skills and user
interface
⚫ Integrates with KDE Connect out of the box
− Pairing
− Playback control via MPRIS
OVOS Shell Demo
⚫ Lets see some of the things the shell can do
OCP Multimedia Framework
⚫ OVOS Common Play is a full-fledged voice media player
framework handling voice integration and playback
functionality for multimedia related skills
⚫ OCP handles all queries related to media playback such
as “play a song”, “play a video”, “play anything”
⚫ OCP framework can be extended by multiple media
providers and uses fuzzy matching to find the best
results to your media query
OCP Demo
⚫ Lets see the multimedia framework in action
How is ovos different internally
Mycroft Core OpenVoice Core
⚫ Micro / plugin based
⚫ Monolithic architecture architecture
Supports multiple STT
Supports limited stt, tts
⚫
⚫ providers: Google STT, VOSK,
engines hardcoded in Conqui-AI STT, more
mycroft-core ⚫ Supports Mimic, Mimic 1,
Mimic 2, Mimic 3, Larynx, Glow,
Tied up to Mycroft’s Pico, Mary TTS, Conqui-AI TTS
engines
⚫
selene backend
⚫ Local/Offline by default,
Mycroft’s selene backend
optional
OVOS Plugin Manager
⚫ OVOS Plugin Manager can be used to search, install, load and create plugins for
the OpenVoiceOS ecosystem
⚫ Every engine is a plugin in the ovos world
⚫ Types of plugins supported:
− Wake Word Engine Plugins
⚫ Precise, Precise-Lite, Snowboy, Pocketsphinx, Nyumaya, Hotkeys, Vosk...
− Speech To Text Engine Plugins
⚫ Selene, Vosk, Pocketsphinx, Google, Self-Hosted HTTPS Service...
− Text To Speech Engine Plugins
⚫ Mimic 1/2/3, Pico, Polly TTS, Google, Mary TTS, Conqui-TTS, Softcala, Cataron, Cotovia,
ResponsiveVoice...
− Intent Engine Plugins
⚫ Adapt, Padaos, Padatious, Regex..
− Hardware Abstraction Plugins
Stack as a framework
⚫ OVOS incorporated into your own project
⚫ Utilize a subset of the services and plugins to meet
the product requirements
⚫ Connect up your native QT Applications to service a
voice interface using the libqmycroft library
Stack as a framework demo
⚫ OVOS integrated into Plasma Bigscreen to serve voice
applications and voice queries
Stack as an operating system
⚫ OVOS on bare metal as an embedded system
⚫ Smartspeakers, Voice Satellites, Magic Mirrors
⚫ Utilizes plugin based hardware abstraction layer
− Mic Array and Audio Board Plugins
− Network Management Plugins
− Display Management Plugins
− GPSD Plugins
Stack as an operating system demo
⚫ OVOS running our embedded shell and packages as a
smartspeaker
Targeting All Platforms
⚫ Embedded headless device or single board computers with
screens to build a do it yourself smart speaker, Openvoice OS
is the open community playground for all platforms to come
and experiment at
⚫ You can expirement with us on:
− Buildroot Based Image for Mark-2 Dev Kit
− Manjaro Arch Based Image for Mark-2 Dev Kit
− Manjaro Arch Based Image for Respeaker based HW
− Manjaro Arch Based Plasma Bigscreen Images
Meet Our Team
⚫ Open Voice OS is a community powered project with contributors
from various parts of the globe working in various industries
⚫ The project apart from being community powered also consist of
lead maintainers for various aspects of the framework
⚫ Lead Maintainers:
⚫ Casimiro Ferreira ⚫ Daniel McKnight ⚫ Peter Steenbergen (J1nx) ⚫ Aditya Mehra (Aix) ⚫ Parker Seaman
(Jarbas) (Strongthany)
⚫ (Daniel) ⚫ Buildroot and ⚫ OVOS Shell and
⚫ Community mycroft-core Community and
developer since 2017 Community mycroft- OVOS Packaging Maintainer OVOS Core GUI Maintainer
⚫
⚫
core developer since
⚫ ⚫
Marketing Manager
Core Maintainer / Dev 2018
⚫
Embedded and Audio Hardware OVOS API Service Maintainer
⚫
Bring Up
⚫
⚫ Documentation and
⚫ PHAL Maintainer ⚫ Core Maintainer / Dev
⚫ Skills UI and more Developer Relations
⚫ Image Build Services
Thank You!
Find us on:
https://www.openvoiceos.com
https://github.com/OpenVoiceOS
Matrix channel links on the website :)
