DiScribe

Introduction

Hi, we are all UBC students who took CPSC 319 Software Engineering Project in 2019 Winter Term 1. This is the code repository for a proof of concept project sponsored by HSBC Canada.

We built a powerful bot that joins phone calls or web meetings, record, transcribe, and send the meeting transcription texts to your email.

The bot recognizes your content of speech and who you are correctly, with time stamps and name of the person.

And YES. This whole process runs silently and autonomously, handling multiple meetings concurrently with no issue.

Feel free to read more about our project in our complete Wiki.

Possible Result

[Happy Tree]	  0:0:0	  We are the workplace futurists. We are happy to see you.

[Excited Duck]	  0:0:8	  Me too. What a great day.

[Amazed Cow]	  0:0:11  So am I. Good bye.

Platforms we support for meeting scheduling

• Cisco Webex Meeting
• MS Outlook
• DiScribe Web meeting scheduling.

Main Components

Website Component

• Allows unregistered users to register in the voice print database with their email
• (stretch goal) Able to save meeting informations into the Database
• (stretch goal) Allows users to manage meeting recordings and their corresponding transcription results
• (stretch goal) Allows users to update voice print samples, their email addresses (or add alternative ones)
• (stretch goal) Warns users when audio input for voice registration isn't length sufficient (We implemented with a count-down and a display of the length of audio recorded)

Scheduling Component

• Able to retrieve meeting attendees email Lists
• Check if any attendees aren't registered in the voice print database
• Sends email to unregistered attendees with link to voice print registration website
• Bot automatically Executes the following processes when meeting time is near

Call-in Component

• Parses meeting informations
• Dials into the meeting on time

Recording Component

• Meeting finishes. Saves Recording into Cloud
• Downloads in correct Format

Transcribing Component

• Starts the Transcribing process upon the recording finishes downloading
• Transcribes the meeting recording into text
• Recognizes the speakers and labelling them
• Outputs the formatted text file
• (stretch goal) Speaker recognition with n > 10 meeting participants

Emailing Component

• Sends the Email to meeting attendees upon transcription result is received

unchecked boxes indicates incompleteness