Skip to content

HACS custom integration for using Whisper speech-to-text (OpenAI or GroqCloud) API in the Assist pipeline, reducing the workload on the Home Assistant server.

License

Notifications You must be signed in to change notification settings

fabio-garavini/ha-openai-whisper-stt-api

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

31 Commits
 
 
 
 
 
 
 
 

Repository files navigation

Whisper STT Cloud API integration for Home Assistant 🏠🎙️

This HA custom integration lets you use any compatible OpenAI API (OpenAI, GroqCloud, others coming ...) for computing speech-to-text in cloud, reducing workload on Home Assistant server.

Sources

  • OpenAI
  • GroqCloud
  • others coming ...

OpenAI

Requirements 📖

  • An OpenAI account 👤 --> You can create one here
  • An API Key 🔑 --> You can generate one here

Models

  • whisper-1 - At the moment is the only model available, despite the name this is the whisper-large-v2 model

GroqCloud

Requirements 📖

  • An GroqCloud account 👤 --> You can create one here
  • An API Key 🔑 --> You can generate one here

Models

Currently all GroqCloud Whisper models are free up to 28800 audio seconds per day!

  • whisper-large-v3
  • distil-whisper-large-v3-en - optimzed version of whisper-large-v3 only for english language

How to install ⚙️

Before configuring the integration you must first install the custom_integration. You can do it through HACS or manually

HACS ✨

  1. Addthis repository to your HACS repositories:

    • Click on this link ⤵️

      Add Repository to HACS

    • Or copy this url ⤵️ and paste into your HACS custom repostories

      https://github.com/fabio-garavini/ha-openai-whisper-stt-api
  2. Install 💻 the OpenAI Whisper Cloud integration

  3. Restart 🔁 Home Assistant

Manual Install ⌨️

  1. Download this repository
  2. Copy everything inside the custom_components folder into your Home Assistant's custom_components folder.
  3. Restart Home Assistant

Configuration 🔧

These are the parameters that you can configure:

  • api_key: (Required) api key
  • model: (Required) Check your source API
  • temperature: (Optional) Sampling temperature between 0 and 1. Default 0
  • prompt: (Optional) Can be used to improve speech recognition of words or even names. Default ""
    You have to provide a list of words or names separated by a comma ,
    Example: "open, close, Chat GPT-3, DALL·E".

Now you can set it up through your Home Assistant Dashboard (YAML configuration not supported).

Home Assistant Dashboard 💻

  • Configure the integration by clicking here ⤵️

    Add Repository to HACS

  • Or navigate to your Devices & services page and click + Add Integration

About

HACS custom integration for using Whisper speech-to-text (OpenAI or GroqCloud) API in the Assist pipeline, reducing the workload on the Home Assistant server.

Topics

Resources

License

Stars

Watchers

Forks

Packages

No packages published

Languages