Inspiration

We say like a lot. Like a lot. Public speaking is scary. We all get super nervous when its our turn to present our speech in class.

Or we have an interview. Or anytime were told to speak in front of anyone. When we do public speaking we tend to say “ I mean, so like, yeah, basically, like you know” Well it’s time we find a way that we stop using these filler words because speech and communication is important in every field in life.

What it does

First you are going to record your voice. You can either present your own prepared speech or you can ask for some prompts to improvise to and this will improve your impromptu speaking skills. The program will give you a random prompt As a result, the program will return 1. The filler words 2. How many times these filler words were used and 3. Your level of preparedness

How we built it

Some Technologies that we used is that

  1. We used Python Speech Recognition Library, With this library, we were able to record the audio from a computer microphone
  2. Google Speech API. Using Google Speech API, we converted the audio into the text format , and lastly, we parsed and analyzed the text to look for any filler words and to count how many of each filler word was used to determine the preparedness.

Challenges we ran into

The audio listener wasn't working as we needed to download PyAudio and we had issues with that. We had to install HomeBrew to help us install portaudio(laptop microphone) first followed by PyAudio.

Accomplishments that we're proud of

We're proud of brainstorming ideas and executing this one. We are all speech and debate kids and so we can really relate to this topic and can see a large-scale impact.

What we learned

we learned so much about speech APIs and also researched sentiment analysis and the challenges of working with audio

What's next for FillerBuster

We want to convert this into a mobile app so that we can use it on the go!!! Some other expansions could be to analyze the tone of the voice (whether it's dull or monotonous or exciting)

Built With

Share this project:

Updates