2 Cloud voice recognition
With Groq
Whisper-large-v3-turbo
This model provided by Groq is IGOOR's default model because:
- Already integrated into Groq's offering through its API;
- it is the de facto standard for voice recognition, used worldwide and supporting a large number of languages with good quality;
- the price is very competitive compared to the market;
- it is very fast.
Whisper-large
As an alternative, in the Whisper extension settings, you can choose the Whisper-Large-v3 model, which is:
- Slightly more expensive;
- slower;
- slightly more accurate.

With Mistral
Currently, in the IGOOR software, the best transcription quality, at least in French, is achieved with the voxtral-mini-transcribe model developed by Mistral. You can use it, with pay-as-you-go pricing, through Mistral's AI Studio.
If you opt for Mistral, you must therefore have an account on Mistral's AI Studio. If you haven't already created one:
Once you have an API key, go to:
Settings > Home > Voice Recognition > Configure voice recognition

On the page that opens, choose Mistral instead of Groq as the provider, and insert your Mistral API key.
