GithubHelp home page GithubHelp logo

mkiol / snipek Goto Github PK

View Code? Open in Web Editor NEW
6.0 3.0 3.0 19.57 MB

Voice assistant for Sailfish OS based on Snips software

License: Mozilla Public License 2.0

CMake 2.27% C 75.96% C++ 8.33% QML 3.52% QMake 0.52% HTML 0.75% Makefile 1.01% Batchfile 0.02% Python 5.31% Shell 2.30% Dockerfile 0.01%
snips sailfishos

snipek's Introduction

Snipek

Voice assistant for Sailfish OS based on Snips software

Important Notice. Staring from January 31, 2020 Snips technology is not longer publicly available. It means that Snipek further development will be very challenging and most likely not possible any more.

Description

Snipek is a Sailfish OS app that provides remote microphone and speaker capability for Snips voice system.

Snips is a voice platform that doesn’t require external cloud service. All processing (including voice) is entirely done locally, on a device where Snips is installed. To make Snipek app works you have to separately install Snips on your Sailfish OS phone or on any another device in your local network. More information about Snips installation options are in Snips installation section.

Snipek app connects to Snips as an additional audio server. It sends audio stream captured from the microphone and plays audio data received from Snips. It also provides few built-in skills that enable phone control with the voice commands. Currently implemented skills are described in Snipek built-in skills section. Possible use cases for Snipek are outlined in Use cases section.

Snipek does not provide Snips software. Snips components should be installed separately on a Sailfish OS phone or different machine (e.g. Raspberry Pi, any Debian based computer).

Use cases

Snipek can be used in several configuration options. Here are few examples:

  1. Snipek is used as a primary mic/speaker for Snips installed on another computer in your local network. Computer where Snips is installed may not have any mic/speaker connected.
  2. Snipek is used as secondary mic/speaker for Snips. Computer where Snips is installed may have mic/speaker connected but it also listens voice commands from Snipek.
  3. Snipek is used as a mic/speaker for Snips and as a voice assistant (voice commands) that let you control your phone. Snips software as well as Snipek assistant are installed on another computer in your local network.
  4. Snipek is used as a pure off-line voice assistant. Snips software and Snipek assistant are installed on a Sailfish OS phone.

Snipek built-in skills

Skills are the capabilities of voice assistant i.e. the things that assistant can do with a voice command.

Following skills are implemented:

Name Skill behaviour Intents Trigger phrase (examples)
Date & Time Reads current time or date. getTime, getDate What's the time? What's the date?
Call History Reads events from call history e.g. all missed calls. getCalls, getMissedCalls Read today's calls. Read missed calls since yesterday.

More skills will be added in the future...

Snips installation

According Snips documentation, Snips officially supports following platforms:

and unofficially (with some hack):

Keep in mind that Snips is not an open source software. The use of Snips is is governed by Snips Terms of Use. The source code is not publicly available. Snips company publishes only binaries for ARM and x86_64.

Snips installation on Raspberry Pi

This is the easiest option because Raspberry Pi is officially supported platform and very well documented. The guide how to install Snips on Raspberry Pi is available here.

Remember to install Snipek assistant if you want to use Snipek voice commands.

Snips installation on Debian

Debian installation is also supported by Snips but official documentation is less clear. Here is quick step-by-step guide for Snips installation on a fresh Debian 9 (stretch) system:

  1. Make sure that your Debian machine is connected to local network and its IP address is "visible" for your phone.

  2. Enable "non-free" packages in /etc/apt/sources.list. For details see Debian guide.

  3. Add Snips repository.

    # apt install dirmngr apt-transport-https
    # bash -c 'echo "deb https://debian.snips.ai/stretch stable main" > /etc/apt/sources.list.d/snips.list'
    # apt-key adv --fetch-keys  https://debian.snips.ai/5FFCD0DEB5BA45CD.pub
    # apt update
    
  4. Install Mosquitto MQTT broker.

    # apt install mosquitto
    
  5. Install essential Snips packages.

    # apt install snips-platform-voice snips-tts snips-watch
    
  6. Download and install Snipek assistant file (skip this step and install your own assistant if you don't want to use Snipek built-in skills or want to create the assistant with Snipek Intents from the console).

    $ wget https://github.com/mkiol/Snipek/raw/master/assistant/assistant_en.zip
    # apt-get install unzip
    # mkdir -p /usr/share/snips
    # unzip assistant_en.zip -d /usr/share/snips
    
  7. Start Snips and check if all components are active.

    # systemctl start snips-*
    # systemctl status snips-*
    
  8. Start Snipek app on your phone, configure IP address and port of MQTT broker (default MQTT port is 1883) and connect.

  9. Say "Hey Snips" (default wake word) and check if Snipek reacts.

Snips installation on Sailfish OS

Snips is not an open source software and only binaries for ARM and x86_64 are available. There is no x86 build (32 bit) therefore installation on Jolla Tablet and any other non-ARM Sailfish OS devices is not possible.

To make Snips setup process as easy as possible, Snipek app provides two bash scripts:

  • snips_download.sh - downloads all needed binaries from Snips/Raspbian/Debian reposotories and downloads Snipek assistant
  • snips_start.sh - starts/stops Snips components

After Snipek app installation, both scripts are in /usr/share/harbour-snipek/snips.

Download (snips_download.sh) must be executed mannually from terminal on SFOS device. Starting/stopping (snips_start.sh) is managed via Snipek app, so there is no need to use that script mannually.

To install Snips software and Snipek assistant in the default directory (/home/nemo/.cache/harbour-snipek/harbour-snipek/snips), run following command from terminal on SFOS device:

$ /usr/share/harbour-snipek/snips/snips_download.sh

Script might prompt you that ar or/and curl are missing. If these commands are not installed already, you should install them and run the snips_download.sh once again:

$ devel-su pkcon install curl binutils
$ /usr/share/harbour-snipek/snips/snips_download.sh

Here are other examples how scripts can be used:

# Display usage help for snips_download.sh:
$ snips_download.sh -h

# Download Snips binaries (on SFOS):
$ snips_download.sh

# Check if all needed files exist:
$ snips_download.sh -c

# Display usage help for snips_start.sh:
$ snips_start.sh -h

# Start Snips:
$ snips_start.sh

# Start Snips in a debug mode:
$ snips_start.sh -v

# Stop Snips:
$ snips_start.sh -k

# Check if Snips is running:
$ snips_start.sh -c

Snipek assistant installation

Snipek assistant provides intents for Snipek built-in skills. Intents are defined for particular language. Currently only English is supported.

There are two ways to install Snipek intents:

  1. Download already created Snipek assistant file and unpack it on the computer where Snips is installed. This assistant only contains intents for built-in skills.

    To download and install on Debian/Raspbian, execute following commands:

    $ wget https://github.com/mkiol/Snipek/raw/master/assistant/assistant_en.zip
    # mkdir -p /usr/share/snips
    # unzip assistant_en.zip -d /usr/share/snips
    

    To download Snips binaries and assistant on Sailfish OS use snips_download.sh script:

    $ /usr/share/harbour-snipek/snips/snips_download.sh
    
  2. Alternatively, create your own assistant with Snips console and add Snipek intents from the store. Snipek intents are published here. If you decide to fork Snipek intents, most likely namespace will change (e.g. muki:getTime to userX:getTime). You can update the namespace to new one on Snipek app settings page.

On Sailfish OS, assistant and Snips binaries are installed in /home/nemo/.cache/harbour-snipek/harbour-snipek/snips directory.

Languages support

Snips supports following languages:

  • German
  • English
  • Spanish
  • French
  • Italian
  • Japanese
  • Portuguese (Brazil)

When Snipek is used only as remote mic/speaker (without built-in skills), it's language-agnostic. It just forwards and receives audio samples to/from Snips.

When Snipek is used for voice commands that let you control your phone, language matters. Currently built-in skills and Snipek assistant support only English language.

Downloads

Binary packages for Sailfish OS can be downloaded from OpenRepos and from the official Jolla Store app.

License

Snipek is developed as an open source project under Mozilla Public License Version 2.0.

The use of Snips assistant file is governed by Snips Terms of Use.

snipek's People

Contributors

carlosgonz0 avatar mkiol avatar

Stargazers

 avatar  avatar  avatar  avatar  avatar  avatar

Watchers

 avatar  avatar  avatar

snipek's Issues

Doesn't work on Jolla 1 phone

Right now, Snipek doesn't work well on Jolla 1 phone. There is a huge delay between voice command and response from Snip system.

Recommend Projects

  • React photo React

    A declarative, efficient, and flexible JavaScript library for building user interfaces.

  • Vue.js photo Vue.js

    🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.

  • Typescript photo Typescript

    TypeScript is a superset of JavaScript that compiles to clean JavaScript output.

  • TensorFlow photo TensorFlow

    An Open Source Machine Learning Framework for Everyone

  • Django photo Django

    The Web framework for perfectionists with deadlines.

  • D3 photo D3

    Bring data to life with SVG, Canvas and HTML. 📊📈🎉

Recommend Topics

  • javascript

    JavaScript (JS) is a lightweight interpreted programming language with first-class functions.

  • web

    Some thing interesting about web. New door for the world.

  • server

    A server is a program made to process requests and deliver data to clients.

  • Machine learning

    Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.

  • Game

    Some thing interesting about game, make everyone happy.

Recommend Org

  • Facebook photo Facebook

    We are working to build community through open source technology. NB: members must have two-factor auth.

  • Microsoft photo Microsoft

    Open source projects and samples from Microsoft.

  • Google photo Google

    Google ❤️ Open Source for everyone.

  • D3 photo D3

    Data-Driven Documents codes.