Please enable JavaScript.
Coggle requires JavaScript to display documents.
FAQs, Live Video Caption - Coggle Diagram
FAQs
Without GPT3
store them QAs in an index in Algolia or MongoDB
search the index given the user question as the query
gather a list of QAs for each company
example:
https://www.mcdonalds.com/gb/en-gb/help/faq.html
A common list of QAs for all the companies [
[Using GPT3]
requirements
: - a list of QAs, both general and specific to a company, preferably a large list
actions
gathering the QAs
Finetuning a GPT3 model based on the QAs
getting users questions
answering the question using the fine-tuned model
Each company has a separate set of QAs [Using GPT3]
requirements: - a QA list for each company
actions
gathering company specific QAs [it needs a dashboard for insert/delete/edit for QAs for each company]
storing QAs in DB for each company
accessing QAs given company_id
answering using GPT3
gettinng a question from a user
retriieving QA examples from DB
Live Video Caption
Actions
buy google speech-to-text api credentials
extracting audio stream from video stream
extracting captions using Google speech-to-text (doc:
https://cloud.google.com/speech-to-text/docs/streaming-recognize
)
embedding the caption stream into the video stream
Requirements
being able to extract the audio stream using Agora
web SDK
Being able to embed the captions in to the video stream using Agora web SDK