Watercooler chat

text-to-speech
#1

The Balena team is remote-first, and almost all of us are working remotely. That has a lot of upsides, but there are also some downsides, such as most communication going on in the team chat. Thought I put together a small project that makes it feel more like having a watercooler discussion going on nearby, to have a more “lively” environment even for those who are remote. Not sure if I’ve succeeded with the environment, but here’s the project nonetheless :stuck_out_tongue_winking_eye:

A demo of the result is in this video. The speakers are connected to a Raspberry Pi 3, that runs the project, while in the left corner there’s my laptop and the regular chat notifications. When there’s a message (and you listen to the video with voice enabled :loud_sound: ), you will hear most messages read out:

The voice is pretty robotic, and would have to tune it for sure to be less creepy :space_invader: but the basic concept seems to work with our chat system. When recorded the video it wasn’t a very busy time, it will be interesting to keep it running more in the future.

The code is here:

It’s using a node.js library to talk to our chat system, and uses Festival for speech synthesis.

What do you think? Would your own team like a robotic watercooler? :robot:

2 Likes
#4

Cool stuff @imrehg :smiley: