The pandemic has transformed how people communicate forever. Most in-person meetings are now just hour-long video calls. But at its core, the art of conversation is still the same. We believe that the key to making a great product is to build something you need. So, we started by listing all the problems we face daily while communicating.
- We tend to hesitate while speaking.
- We lack confidence.
- We overuse common words while speaking.
- Grammatical Errors while speaking.
Kit attempts to solve all the above issues effectively.
What it does
Simply put, Kit is a multipurpose video calling platform, which helps people improve their communication skills. How do we do it? Let me explain with an example.
Video Conferencing Platform
Suppose you have a debate competition coming up in the next week.
How do you prepare for it? Enter, Kit.
Visit our website and create a new meeting. While doing so, select your interest area. You can choose from a wide range of options like Environment, Politics, etc.
After picking something, send the invite link to one of your friends.
Wait for them to join. Kit presents a question to both of you based on your interest. You can choose to speak in favour or against the motion.
The debate begins.
Both of you take turns and say what you have to say.
The meeting ends when both of you have nothing left to say.
AWS Transcribe transcribes the meeting audio for both the participants.
The transcribed text from both the participants' audio files is fed into multiple state-of-the-art Natural Langauge Processing algorithms to get meaningful insights. These insights are captured in a report and are mailed to the users after the meeting ends. They can view these reports on the web application. Every report has -
Sentiment Analysis - What was the tone behind what you said? Positive or Negative?
Behavioural Analysis - What was your state of mind during the meeting
Emotional Analysis - How did the other person react to what you said?
Word Cloud - What words did you use?
Most Used Words - What are the words you use the most? Are there any alternate words?
Transcript - What did you say during the meeting?
Grammar Checker Tool - What are the common mistakes you make while speaking?
Sentiment Analysis is a basic NLP technique that gives you a more human-like understanding of the text. Emotional & Behavioral traits provide insights into your personality type by capturing the specific emotions you project. This can help individuals answer questions like how to be more likeable.Word Cloud is a data visualization technique used for representing text data in which the size of each word indicates its frequency or importance. The Most Used Words feature helps you expand your vocabulary. Studies have shown that the lack of proper communication is one of the reasons why most candidates get rejected in a job interview. While speaking, using poor grammar can lead to ambiguity and misinterpretation. So, the question that now arises is how do we solve this problem? Kit transcribes every word you speak in the meeting and stores it for inspection. The built-in AI-powered Grammar Checker Tool not only points out all the mistakes you make while speaking but also finds a suitable substitute for all the errors.
How we built it
- The front end uses React & Redux.
- Nodejs & PostgreSQL to store the data.
- Python & Flask for Natural Langauge Processing & Artificial Intelligence.
- Video calling using the Dolby.io Communications APIs
- AWS transcribe to transcribe recordings
- For the Grammar Checker, we used the Hugging Face Transformers pipeline.
- The NLTK python library to generate the synonyms
- The word cloud python library for making word clouds
- The Sentiments, Emotional traits & Behavioral traits that we get are from Expert.ai's NLP APIs.
Challenges we ran into
- Learning about Dolby's Communications APIs
- Making the demo video
- Setting up Dolby API to work in tandem with AWS Transcribe
- Designing the overall architecture of the application
Accomplishments that we're proud of
What makes us proud is that we were able to come up with something truly novel and revolutionary. The use-cases for Kit are infinite. Kit works for everyone, whether you are a working professional or a student. Our application will help you become a better communicator. In the future, we look forward to rolling out new releases and delivering a better product.
What we learned
Where do I start? We are a small team of college students from India. Every member was working from home due to the current global situation. Despite the difficulties, we managed to work together just fine. This hackathon has strengthened our bond as a team and made us better at collaboration. On the tech side of things, learning Dolby's Communication API's was fun. Building the Grammar Checker Tool required us to have a thorough understanding of Natural Language Processing. And before I forget to mention, we also learned the basics of Premiere Pro & Adobe Spark for making our demo video. We hope you like it!
What's next for Kit
Currently, we focus only on the audio. In the future, we plan to integrate computer vision into the application. Tracking the facial expressions and emotions of the participants will help improve reporting drastically. It will also help us analyze all the gestures made during the call. The reports are only available via the online interface. However, in the next release, the users will have the ability to export them as pdf and use them on the go. Lastly, We would like to thank Dolby for organizing this hackathon.