The landing page of our web interface
One of the most fundamental and yet challenging skills that one has to use in everyday life is communication. The way that others see you is mostly based on the way you can present yourself. School is supposed to prepare us for the challenges that face us in adult life, and it mostly succeeds in terms of providing us with basic knowledge, however it usually fails at delivering us the communication skills we absolutely need to be able to succeed in the modern world.
What it does
Have you ever been practicing for an important presentation in front of the mirror? I believe we all can agree that mirror is not the best judge of your abilities, since its limited to your own perception of your performance. What you need, is an unbiased point of view. This can be provided by our novel approach, which lets you evaluate your oratory skills and opens new ways to improve them.
How we built it
The core of our application is the backend that includes integration with Google Cloud API (NLP module, Speech to text and vice versa, and IBM watson framework) that handles all the logic with audio data. Then on top of it we have built custom modules to extract beautiful insights from the data. For example, we can analyze text complexity from its grammar structure and texts statistics. Furthermore, we can estimate the level of calmness of the person’s speech and even find a character or a real person whose manner of speech is relatively close to the user’s. This is all built using custom data, models and services deployed with the power of Google Cloud.
From the user perspective the communication is as simple as possible. He interacts with a fancy UI with a recorder and gets all the insights directly after his speech is processed.
Our project has many fields of application. It can be used as a hight level API for any other service that needs voice features or it can be used to build a separate business app that solves an exact problem like IELTS speaking part preparation for foreign students.
Challenges we ran into
We a spent a lot of time working on a different project which we ended up discarding.
Accomplishments that we're proud of
We have never before worked on a web development project, and we are proud that we were able to take it to the end, even though we lost a lot of time working on another project.
What we learned
Everything we now know about web development:)
What's next for SpeechCoach.
We didn't have enough time to finish all of the features and implement all of our ideas, which we know would greatly contribute to user experience. For example, we absolutely need to change our neural network structure and train it further using more data, since it was extremely challenging to find a dataset as it was.