How Rev AI Helped Create a Property-Searching App

University of Texas Austin computer science major Ethan Houston used the Rev AI advanced speech recognition API as part of his team’s project at HackTX. After perusing the hackathon sponsors’ different challenges, Ethan’s team decided to work on a property-searching app, Birdhouse, that incorporated the Rev AI speech-to-text API. Birdhouse users would be able to use a natural language search for their ideal properties.

Participating in HackTX

HackTX is an annual, 24-hour hackathon held at the University of Texas at Austin. Open to all UT Austin students over age 18, this event provides workshops and mentorship to hackers at all skill levels. The experience also includes mini-events, challenges, and prizes.

Ethan Houston, a UT Austin freshman majoring in computer science, enjoys many aspects of web development. So far, his specialty focuses on mobile applications. As a student and veteran hackathon participant, he is always on the lookout for new opportunities to build on his knowledge.

Combining Automatic Speech Recognition and Property Searches

Several sponsors set up booths at HackTX, offering participants the chance to win prizes in exchange for tackling specific challenges. Ethan and his three teammates took 1-2 hours to brainstorm after collecting ideas from sponsors, including Rev AI’s challenge to incorporate speech-to-text.

Ethan’s team decided to develop Birhouse, a web application centered around real estate using Rev AI. The idea was to have a simple app in which the user could speak into a device to provide search criteria and receive a list of matching properties. This might include the type of property, their desired city, and the preferred price range.

Next came the challenge of creating an app that would use Rev AI’s API to do just that.

Using Rev AI to Build the “Birdhouse” App

To fulfill its purpose, Ethan’s app needed Rev AI to convert the user’s spoken audio to natural text. This text would then be parsed and interpreted to extract the key features, such as city and price range. From there, the app would invoke the Zillow and Google Maps APIs to curate and display a list of properties.

Ethan’s team hosted their project on Google App Engine, a web framework and cloud-computing platform. They used Python’s Flask framework to create a simple web application with one or two front-facing pages. These pages were intended to be simple, only displaying a microphone symbol to click on before speaking the property keywords.

They used the Javascript browser API to capture the user’s audio and compress it before sending the file to Ethan’s API. That API would store the raw audio and submit it to Rev AI for transcription.

One of Ethan’s teammates, Stefan deBruyn, wrote a custom text interpreter. The interpreter would “read” the long text string returned by Rev AI and extract essential data, such as the city or budget. The Birdhouse app could then send a request to Zillow’s free API with the user’s parameters.

Zillow would return a list of matching properties and their latitude and longitude values. Ethan’s team then used the Google Maps API to deliver the final results as map with pins where the matching properties were located.

Ethan’s team spent 18-20 hours at the hackathon building Birdhouse, plus 1-2 hours brainstorming and occasional breaks for rest or testing.

“We had four different people work on four different parts, and then, at the end, we just had to do a little bit of connecting,” Ethan explains. “It was great fun.”

Rev AI Provides Straightforward Learning Experience

Although unfamiliar with Rev AI before the hackathon, Ethan and his team were interested in the challenge set forth.

“When we came in, we weren’t expecting to do a speech-to-text thing,” Ethan says. “We’re thankful they were there because that gave us another thing to learn about and use.”

The HackTX team found the Rev AI documentation straightforward to work with. Additionally, they were able to speak with company representatives at the hackathon about questions and setup errors.

Another strength for Rev AI was the accuracy with which it transcribed spoken words into text. “We never even really factored in the chance that the text we’re getting out of the API might have typos or errors,” Ethan says.

While the Birdhouse case is not as complicated since only one person speaks at a time, this accuracy is especially important when distinguishing words spoken by two or more individuals.

Using Rev AI for Additional Future Projects

Since using Rev AI for Birdhouse, Ethan has already thought of other ways to use the speech-to-text API. Besides being a UT student, Ethan is also a co-founder of Watshout. This San Francisco-based startup aims to create a next-generation smartwatch tailored toward athletics.

Before releasing their product, Watshout is working to create a mobile app to demonstrate its capabilities. This will focus primarily on safety, a common concern for anyone who exercises away from home or alone. The mobile app will include a tracking feature that allows the runner to notify designated people while they are using the app.

Rev AI could be useful for this app as a way to give the software voice commands or send a message. This would eliminate the need to stop and type, interrupting their workout and potentially compromising their safety.

Try Rev.ai for Yourself

Rev AI is an advanced speech recognition API from the makers of the transcription services Temi and Rev.com. Power your application with best-in-class proprietary speech models. Plus, real-time transcription is coming soon!

Rev AI was built by leading speech-recognition experts using millions of hours of accurate, human-transcribed content and data. Try Rev AI for free today and see what’s new in speech-recognition technology.