Speech Search

Inspiration

After seeing Assembly AI's speech-to-text API, and how it had timestamps, we got the idea to make an app to help those who need to find the right section of an audio file when there is no closed captioning readily available. (Video is still in the works).

What it does

Once a URL to a valid audio file is provided, AssemblyAI's API runs and returns a list of words in the file. Our app, then, provides a search bar to isolate keywords that you wish to find, and navigate directly to any resulting timestamp where such a keyword is spoken.

How we built it

For front-end, heavy HTML and CSS were used. They were then brought into React, with some HTML becoming components, and others staying raw HTML that got passed in throughout various functions. The API calls were made with the JS library Axios.

Challenges we ran into

Getting videos to work with the API proved to be difficult, so we worked on ensuring audio worked. Also, uploading files to the server was not a problem, but the API did not play friendly with accessing the transcript of the uploaded file.

Accomplishments that we're proud of

Building a sleek CSS and working with the API!

What we learned

Building a web app definitely isn't a walk in the park, especially when combining front-end and back-end together after hours spent on each part independently.

What's next for Speech Search

Next, we hope to get video integration as well as local file uploads finished. We're partway through each as of the end of the hackathon.

Built With

• assemblyai
• css
• html5
• javascript