Steaming STT using AI model

[jacko6]

Has anyone successfully added streaming (i.e. near realtime) speech to text using an API interface to any of the recognised AI speech models?

I know you can use the Win+H key but I need much higher accuracy plus the additional features/parameters that the speech AI models provide.

Example providers are AssemblyAI, Speechmatics, DeepGram...

Ideally I want to do it all in VBA, via websockets etc. I've asked the OpenAI and Claude to code it but they (and I) can't quite get it working.

 

[arnelgp]

you may start with this one:

Integrating speech recognition with VBA

Blog: Integrating speech recognition with VBA

www.wiseowl.co.uk

 

[xavier.batlle]

And continue with this other one:
https://www.devhut.net/adding-speech-recognition-to-your-application/

 

[isladogs]

And then look at this one

Dictation in Access

Although not officially supported, it is possible to use speech to text (dictation) to input data in Access. This article explains how to do so.

isladogs.co.uk

 

[jacko6]

Thanks, I was aware of those options but unfortunately they all come up well short of AI model capabilities, which is why I am wanting to use a hosted AI speech model via an API.

For example the AI speech models support countless languages, accents, locale spelling differences, key terms, industry specific models such as medical, confidence temperatures etc.

Even the Win+H key option which uses AI gives poor results compared to configuring a speech model via the API options.

I'll keep at it...

 

[Gasman]

Well if you do ever find what you are looking for, please post back your results.