If we are talking about the same package speech_to_text, maybe there's no way. It wraps all native systems voice recognition, so there's not a direct way to customize the dictionary. For example:

https://developer.android.com/reference/kotlin/android/speech/SpeechRecognizer https://developer.apple.com/documentation/speech

Analyzing both Android and iOS APIs, they are very limited to download languages and in most cases, infering in some results like Android AlternativeSpan, but I don't think that's the case.

I don't know if you are planning to use only local or cloud Speech to Text recognition, but I would recommend Google Speech-to-Text API.

It's available on googleapis package.

I've also searched for some edge models and could only find https://huggingface.co/openai/whisper-small as tiny version, but not sure how "tiny" it is for a mobile phone.