0

I recently started using the RASA framework for developing chatbots. My goal is to create something that once deployed on a cloud VM, it can interface with voice modules so that it can be more easily accessible (no typing required from the user).

Do you have any suggestions as to how I could bridge the RASA bot with the text-to-speech module as well as the speech-to-text (transcription) one? Any help on this would be greatly appreciated. Thanks

Zack
  • 181
  • 1
  • 11

1 Answers1

0

You can use the STT module as a preprocessing step, and the TTS module as a postprocessing step.

  1. Let the STT module process the speech.
  2. Take the output of the STT module (text) and send it to the REST channel of your Rasa server.
  3. Take the REST response (text) and send it to the TTS module.

The more complicated way to integrate these modules into Rasa is to write a Custom component at the top of the pipeline for preprocessing speech, and you have to modify the REST channel to ingest the speech. The TTS can be easily integrated using Custom action to call your TTS module, then the speech is your choice of handling.

Hoai Nam
  • 45
  • 10