#Is there a way to create a custom webhook automation for Faster-Whisper?

1 messages · Page 1 of 1 (latest)

latent matrix
#

Hey all, I currently have Faster-Whisper set up on my local PC via Docker for use with my Home Assistant VPE, which works fairly well. Besides just using Faster-Whisper exclusively for my VPE pipeline however, I would also like to use it in parallel for my other project which transcribes my voice into text.

Therefore, would it be possible to setup a webhook automation that accomplishes this (by receiving voice input and generating text output via HTTP)? I've stumbled across this github issue thread (https://github.com/rhasspy/wyoming-piper/issues/9) on Wyoming-Piper that shows that it is possible to setup such a webhook for Piper, but I'm not sure if there's an equivalent solution for Faster-Whisper. Any advice or suggestions are greatly appreciated, thanks in advance.

(solution for piper below, hoping there is a similar webhook possible for whisper)

description: ""
trigger:
  - platform: webhook
    allowed_methods:
      - POST
      - PUT
    local_only: true
    webhook_id: speak
condition: []
action:
  - service: tts.speak
    target:
      entity_id: tts.piper
    data:
      cache: true
      media_player_entity_id: media_player.snapcast_tts
      message: |
        {{ trigger.json.text }}
mode: queued
GitHub

Documentation doesn't tell me, if its possible to use piper via http requests. I have wyoming piper running in a separate server and home assistant is able to use it just fine. I would like to ...

atomic wagon
#

I'm unclear if your second project uses home assistant? Perhpas Whisper can listen on more than one port and you can differentiate by address.

latent matrix
# atomic wagon I'm unclear if your second project uses home assistant? Perhpas Whisper can list...

Hi thanks for the advice, and yes I would like my second project to use the home assistant wyoming pipeline as well, but I'm having trouble figuring out how to send my voice inputs to the Whisper port (the only solution seems to be setting up a webhook via HA automations which is what I'm stuck on). Fwiw currently my second project is a frontend page on Python where I hold a button to enable push to talk and it would transcribe my voice input into text

atomic wagon
#

Claude says to use either the rest command or a Python script component. He wrote an extensive guide with code. He's usually wrong on the first pass, but I'm sure there is value in his response. I do pay him $20/month, but he probably does the same for free. I don't think you need wyoming

mild rampart
#

I'm sure there's not.

#

And using an AI to help support others is frowned upon.