Skip to content
/ WordWave Public

POC for speech recognition and visualisation

Notifications You must be signed in to change notification settings

pazka/WordWave

Repository files navigation

WordWave

Objective

This project is made for the artwork "Sonus Microform" directed by the artist Alessia Sanna. Multiple clients will display a data visualisation of the word spoken during a conference.

Go watch in production : https://wordwave.hosh.it/data.html

Result

Behaviour

A unique client will register and manage a server. The server will broadcast all word modification to multiple clients which will display those words.

Visualisation :

?dark => Invert Black&White

?min=N => won't display words which occur less than N times

conférence

Admin interface :

image

Recording / Not recording :

image image

Open Data :

On the interface or at urls :

words/current/recorded -> Logs of recorded words
words/current/registered -> Logs or counted words
words/current/count -> total count
words/current/meta -> meta informations
words/current/stop -> excluded word
words/current -> meta + count

Build your own

DEV Requirements :

  • python 3.9
  • docker
  • node
  • gulp
  • gzip

PROD Requirements :

  • docker
  • gzip

Build an image

npm i
gulp

The compressed file is in the dist folder

Deploy the image :

  • Copy and de-zip the build : gzip -d wordwave.tar.gz
  • Load in docker : docker load < wordwave.tar
  • Run in docker : docker run -p <your_port>:80 wordwave
  • To override the default credentials : docker run -e LOGIN=<your_login> -e MDP=<your_mdp> -e secret=<your_secret> -p <your_port>:80 wordwave or docker start WordWave

WIP

TODO :

  • Gradual culling of weak words (7/8 culled for 7000 registered and 2500 counted)

  • Gradual center + darken + Smallest when comming to certain number of words

  • Clean up Render function to incroporate those function in a clean anner

  • Plural test (remove -s)

  • Sonus Microform keeps screen on wake up = 1h

  • Test voice register when screen is hidden = 1h

  • Optimize WordWaveWeb for better perf (P5.js / WebGL / Unity) = 3h

About

POC for speech recognition and visualisation

Resources

Stars

Watchers

Forks

Packages

No packages published