WeBAD = Web Browser Audio Detection/Speech Recording Events API

Hi all,

just to share with you my draft project:

I experimenting how to use the browser as a voice interface for … anything.
Of course in the RHASSPY realms, the browser could act as a perfect “satellite” device/frontend.

Here below a quick video demo of one the voice interactions, I call “continuous listening”:

I appreciate any kind of contribute/discussion. Here or, for general open points, feel free to open discussion topic here: https://github.com/solyarisoftware/WeBAD/discussions



hi, when i run the demo page i get the error ‘Stream generation failed’, what i’m doing wrong?
And how can i use this with rhasspy?

Hi Dariel,

Which browser? Which OS? BTW, I suggest to user firefox!
You read this error on console, right? Which code line?
Do you enabled / have a microphone, right? See by example: https://github.com/cwilso/volume-meter/issues/9

The idea is to use the web browser as voice interface. Audio I/O message could be sent to the Rhasspy base station, no? See: https://github.com/solyarisoftware/webad#architecture

BTW, a :star: on github is welcome :slight_smile:


hi, i change the server in npm from express to the one you suggest and boom, it´s working.

The idea is to use the web browser as voice interface. Audio I/O message could be sent to the Rhasspy base station, no?

Yes, that’s why i have interest in this project, i want to use wall mounted tablet´s to show the dashboard and use the integrated mics to send the voice commands to rhasspy, sorry my ignorance but how can i integrate this to rhasspy?.
I was looking a way to do that and the only info that i found was this topic, thanks for build this.

1 Like
  • The point with Web Audio API is that you you must serve the web page from a server through HTTPS (or running in localhost)

  • So far, I didn’t integrated my project into RHASSPY. The WeBAD is just a component/piece of the puzzle, that supplys javascript events, allowing any user web application (client) to record a speech message and send the audio to a server for any backend processing. As explained in the weBAD docs, there are many way to take the user speech. That project specifically face the “continous listening” mode and few others, just escaping the common “wake word” approach.

    So, as shown in the architecture sketch, one could send the audio recorded on the browser to the RHASSPY base-station to be processed by an ASR. And viceversa, the web app could play an audio message sent from the RHASSPY base-station.