AI for Speaker Tracking & Analysis Python

My startup company has a MVP mobile app that transcribes user’s conversations and does diarisation (Deepgram), sentimental analysis (GPT-4), reflected in scores of speakers, and provides a summary (GPT-4) of conversation.

If it is an audio with 5 speakers, the IndividualRecord page will have 5 such windows with scores.

All outputs from the backend with speakers, scores, summary and full text are stored in the Firestore database.

Current general working principle of this MVP: analysis of INDIVIDUAL, INDEPENDENT audio conversations (meetings, appointments, chats) WITHOUT ANY GENERALIZATION over ALL conversations, speakers.

Desired general working principle of this MVP should be: analysis of SAME SPEAKERS OVER DIFFERENT INDEPENDENT AUDIO.

Full task [login to view URL]

Python

Deep Learning

GPT-3

NLP

Machine Learning (ML)