I’d like to know about Rasa’s character and time limits. I’m building a bot that takes in long voice memos from telegram (upwards of 10-20 minutes of transcript dialog). I understand there’s a 1000 character limit on intent recognition, but, on my local machine, I have a way around that and I’m not sure it’s compatible with RASA.
Let’s say my bot asks questions in a sequence, “What do you think about that problem?” and “Would you like to tell me more?” The user could respond with a 10 minute voice note/transcript to this question and the follow up.
No matter how long the note is, it’s always going to follow up with this or one like it. Now, I know that an NLU would have a hard time parsing a 10 minute transcript. So, I only take the first part of the note for intent recognition and use regex on the rest, I have something like “I’d like to save a note about [A-z]*” to capture all text no matter how long after the user says they’ll save a note.
On my local machine, I also scan the transcript for keywords, like “fail”, to see if the bot should respond with a different list of questions.
Right now, I have go through google cloud to transcribe the audio and return it to the bot as an intent.
Will there be any problems trying to do this with RASA? I’m still considering using it and want to know if it can handle my features.
Thank you and let me know if i can make this question any better,