webXB.

A voice-interview chatbot. Respondents open a link and talk: the bot asks each question out loud, transcribes the spoken answer, checks it fits what the question expects (converting it if it doesn't), and decides what to ask next — the next question in a scripted tree, a conditional branch, or a follow-up it writes from the conversation so far.

Two repos: a Next.js chat front end with a microphone recorder, live sound visualiser, countdown prompts and rating or picture-choice answers, and a Node/Express conversation API that runs the question trees, speech and transcription, stores every conversation in MongoDB and exports them as CSV for analysis. Each tree carries its own bot name, branding and redirect, and the interface runs in English and Hindi.