harlantwood / ai-interviewer-cli

AI Interviewer CLI

TL;DR: AI interviews you, on whatever topics you want, creating audio archives and transcripts.

NOTE!

I am mostly working now on the web version instead of this CLI version, see https://github.com/harlantwood/ai-interviewer

Motivation

This project is primarily to scratch my own itch: have an AI interview me, to get me to talk about topics I want to share with the world. Also to explore AI a bit, and hopefully release something useful for others as well as myself.

Quickstart

• install llm
• install ospeak

cp .env.example .env   # and edit as needed
pip3 install -r requirements.txt
./interview.py

Features

Initial release will be purely CLI based workflow - no GUI, nerds only :)

• Create context (YAML config file):
  • Description of interviewee
  • List of topics to talk about
  • System prompt describing interviewer characteristics
• AI Interviewer asks question about a topic
  • Generated by LLM based on context above
• Speak question out loud to interviewee
  • Ideally record audio, although we can reconstruct it later (in higher quality) from the text
• Save question as text
• Record audio of interviewee answering question
• Transcribe interviewee audio to text
  • realtime or higher quality? probably start with realtime, for better interview flow
  • whisper.cpp?
  • transcribe_demo.py?
• User hits spacebar to indicate answer complete, ready for next question
• Feed text back to interviewer as part of "chat" flow, to generate next question