calling speech to text before bot could generate one result in empty transcriptions

[minhle-vulcan]

Go to DMs with bot, send a voice message, then immediately request speech to text (by clicking on the audio message you just sent). Notice this has to be faster than bot responding, because bot logic is to create a speech to text transcription and save it to matrix if there isn't one. The goal here is to achieve a speech_to_text request on the client side, not done by the bot.

The response of such request will result in an empty transcription.