Wow this is exactly what I had in mind for "opensource tools for speed-to-text". I didn't know it did this too. Thanks a lot!

> voice2json is optimized for:

> Sets of voice commands that are described well by a grammar

> Commands with uncommon words or pronunciations

> Commands or intents that can vary at runtime

Doesn't sound like what you'd want for a generic transcription service.

It supports open-ended transcription too: https://voice2json.org/commands.html#open-transcription

Users have reported good accuracy with the English Deepspeech profile: https://github.com/synesthesiam/voice2json-profiles