RASA Voice bot

Hi All, I am creating speech to text and text to sppech chatbot in RASA and for that I’m following this post( How to build a voice assistant with open source Rasa and Mozilla tools ) As I am stuck with socket.io only. I have all set up ready but the connection is not working as maybe some files are missing or what.

I would like to describe it to you in more detail that all these commands are running fine and after executing it on chrome, After last command npm run serve is not refreshing and staying on the same that jumping ball page and waiting for connection.

And last command python3 -m http.server 8888 is showing my whole project directory page. Is it okay? and after it I’m loading that npm run serve.

It would be greatful If you can lookout to this issue. As I have find that so many people are getting the same issus.

Thanks

If you have any doubts please checkout this YouTube playlist about rasa voice chatbot. This will definitely help you. Thank you

Hi, @skjainmiah. Yes, I checked this, very well explained. But there I’m getiing error…

Check this post and all the links included in it:

Hi @Gehova, Thanks for your response. But in my case I won’t be able to refresh the jumping ball page to voice bot page. I can execute -> rasa train nlu command and from mentioned link i can also execute last 4 server commands and it is also opening in my browser like I’m getting rasa version , in docker I’m getting “quack” message in last command (http server command) I’m getting all project directory there. but after it when I’m trying to refresh the page showing jumping ball( from npm run serve command) it is not refreshing to the voice bot page as it is mentioned in blog. So it is not able to establish connection at the end, my bot is stuck at that jumping ball tab only even after hitting that last command and refreshing the page of jumping ball.

In the credentials.yml:

#socketio_connector.SocketIOInput:
#    bot_message_evt: bot_uttered
#    session_persistence: true
#    user_message_evt: user_uttered

socketio_connector.SocketIOInput:
   user_message_evt: user_uttered
   bot_message_evt: bot_uttered
   session_persistence: true/false

The commented credentials are the ones from the tutorial, replace it with the other. And in socketio_connector.py replace the blueprint function in SocketIOInput with:

def blueprint(self, on_new_message):
    sio = AsyncServer(async_mode="sanic", cors_allowed_origins='*')
    socketio_webhook = SocketBlueprint(
            io, self.socketio_path, "socketio_webhook", __name__
    )

Hey @Gehova, Thanks a lot for your response. Thanks a lot, I have been trying on this for 20 days. And I’m getting the RASA voice Assistant page but If I speak something it is loading only, nothing happened. I would like to attach screen shot of the terminal window and inspect element both. Please find it.

And then do I need to change in nlu, domain and story files to get the custom changes. Right? Again Thanks a lot!!!

hi did you find a solution for this? if yes please tell. im stuck here too

That is the solution. Did you read the post from Neil Stoker that I put in the first response of these thread?

Hi, I got it working. Thanks. Yet the UI is taking too long to respond, sometimes it doesn’t respond at all. help please. The audio files are getting created in the backend, Ui is not loading them or loading them late

In my experience the slow parts is the stt and the tts. Add some prints with timestamps in order to debug that part.

@Gehova

Are you still working on this? Does it make sense to start working on this now since I read that Mozilla software that Rasa uses for speech to text is now in maintenance mode - Mozilla Org Wide Updates: Impacts on Common Voice - Common Voice - Mozilla Discourse

I have made changes as @Gehova said but not getting result as yours, I am still getting “Waiting for connection …” for a long time.

can you help me with this??

Thank you in advance.