https://github.com/fabio-garavini/ha-openai-whisper-stt-api
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.
https://github.com/fabio-garavini/ha-openai-whisper-stt-api
groq home-assistant openai stt whisper
Last synced: 2 months ago
JSON representation
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.
- Host: GitHub
- URL: https://github.com/fabio-garavini/ha-openai-whisper-stt-api
- Owner: fabio-garavini
- License: agpl-3.0
- Created: 2024-03-06T11:11:59.000Z (over 1 year ago)
- Default Branch: main
- Last Pushed: 2025-03-27T14:35:57.000Z (3 months ago)
- Last Synced: 2025-03-27T15:28:50.000Z (3 months ago)
- Topics: groq, home-assistant, openai, stt, whisper
- Language: Python
- Homepage:
- Size: 73.2 KB
- Stars: 37
- Watchers: 1
- Forks: 6
- Open Issues: 1
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# 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*
- *Custom*
## OpenAI
### Requirements 📖
- An OpenAI account 👤 --> You can create one [here](https://platform.openai.com/signup)
- An `API Key` 🔑 --> You can generate one [here](https://platform.openai.com/api-keys)
### Models
- `gpt-4o-mini-transcribe`, `gpt-4o-transcribe` - [Next generation](https://openai.com/index/introducing-our-next-generation-audio-models) OpenAI transcribe models
- `whisper-1` - Despite the name this is the *whisper-large-v2* model
## GroqCloud
### Requirements 📖
- An GroqCloud account 👤 --> You can create one [here](https://console.groq.com/login)
- An `API Key` 🔑 --> You can generate one [here](https://console.groq.com/keys)
### Models
Currently all GroqCloud Whisper models are free up to 28800 audio seconds per day!
- `whisper-large-v3`
- `whisper-large-v3-turbo` - faster version of *whisper-large-v3*
- `distil-whisper-large-v3-en` - optimzed version of *whisper-large-v3* **only for english** language
## Custom
Any other OpenAI compatible api
## How to install ⚙️
Before configuring the integration you must first install the `custom_integration`. You can do it through HACS or manually
### HACS ✨
1. **Add** ➕ [this repository](https://my.home-assistant.io/redirect/hacs_repository/?owner=fabio-garavini&repository=ha-openai-whisper-stt-api&category=integration) to your HACS repositories:
- **Click** on this link ⤵️
[](https://my.home-assistant.io/redirect/hacs_repository/?owner=fabio-garavini&repository=ha-openai-whisper-stt-api&category=integration)
- Or **copy** this url ⤵️ and paste into your HACS custom repostories
```url
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** ⤵️
[](https://my.home-assistant.io/redirect/config_flow_start/?domain=openai_whisper_cloud)
- Or navigate to your `Devices & services` page and click `+ Add Integration`