Intelligence Ears

In the world of interview, million of words are being spoken. Interviewer needs to focus on many things. Would it be better to have extra ear to help both listening and score the interviewee.

Comment

Inspiration

In the world of interview, thousand or million of words are being spoken from both interviewer and interviewee. The work load is not only on writing up and preparing the questions for the interview but to also critically listen to what the interviewee reply back as this can lead to some further specific questions. Would it be even better if there is a tool that help the interviewer listen, analyze and score the interview.

What it does

The user upload the voice recording file and our system will analyze the file and give a score base on the predefine rules. The system return the total score and the words from the key_words that appear and do not appear in the interview.

How we built it

We use the Deepgram's Python library to implement the application. The application performs the business logic in counting the occurrence of each words and return the score.

Challenges we ran into

Some of our team members face the problem of SSL when trying to execute the source code with the Deepgram's API. We try to generate new KEY and many more ways but it does not solve the problem. So, we decided to split the team into two sub-teams. The members that can run the Deepgram continue to work on this and the rest work on the new challenge.

Accomplishments that we're proud of

This is the first time that we ever develop the text recognition related software. Our team is comprise of people from different background such as software engineer and embedded engineer.

What we learned

The most important thing is learning about the text recognition process and also how to use the text recognition API.

What's next for Target wording

The next phase would be to integrate with NLP. Each person has his/her own speaking style and technique. We would use the system to first identify that which person pass the test (he/she cover all the key words) and to further identify the pattern and style which could be used in enhancing the existing set of rules. Not only that it can also be used for the training purpose for the existing employee (this needs to be under GDPR as it is a data that can uniquely identify the interviewee). Thus, we can train the employees to talk critically and concurrently how to answer the expected and unexpected questions.

Built With

Share this project:

Updates