[API Proposal]: Speech to Text Abstrations for Microsoft.Extensions.AI

[sandrohanea]

Background and motivation

It's great that dotnet/extensions is providing abstractions for AI paradigms and a great addition (besides chat and embeddings) will probably be Speech to Text.

This is more of a conversation starter as we need to look what would be the best pattern to accommodate different approaches. I would be happy to contribute with an implementation for https://github.com/sandrohanea/whisper.net (local inference using whisper.cpp)

It should most likely also accommodate: Whisper API (from OpenAI / Azure)
but also https://www.nuget.org/packages/Microsoft.CognitiveServices.Speech

API Proposal

TBD

API Usage

TBD

Alternative Designs

No response

Risks

No response