VoiceVibes uses AI to automate the process of polishing speech

Founder Debra Bond Cancro on the company's algorithms and approach: "The idea of giving people feedback about how others perceive them is what makes us really novel."

Gene editing could be used to fight cancer before long

(Photo by Flickr user socialmediasl444, used under a Creative Commons license)

When it comes to public speaking, Debra Bond Cancro said the words aren’t the only important factor.

“Studies do show that when you’re making an impression,” she said, “the vibe that people get from you is more important than the actual words.”

But while words can be edited endlessly, preparing the way a person will sound can be tricky to approach. There’s a lot to fine tune, such as a speech’s pacing and clarity. Then there are elements like confidence and authenticity that aren’t as easy to quantify — and it’s tough to know how the whole thing will land with an audience.

Cancro, who is the founder and CEO of local startup VoiceVibes, has numbers and data behind those areas. Over several years starting in 2012, the company compiled millions of data points to measure 20 areas that affect how speech will be perceived. With that data, the company trained algorithms to be able to analyze how an audience would perceive the speech. The perception data was rated by humans, and the team calculated the average scores over time.

VoiceVibes offers a web-based app that allows users to practice a speech, and the system then provides feedback based on the 20 areas. The idea is to automate parts of that process of preparation.

“We’re predicting how others will perceive you and what vibe you’re giving off,” she said.

A look at VoiceVibes' dashboard. (Courtesy photo)

A look at VoiceVibes’ dashboard. (Courtesy photo)

It can be adjusted for various scenarios along with speeches, and that’s led to a business focus on providing the product as a tool for companies seeking to train employees. The feedback can be useful to someone who is making a sales pitch, or conducting an interview.

“We’re finding companies are investing so much in the time and effort to craft messages,” Cancro said.

The system can also help in training managers, as well as specific roles. VoiceVibes is working with Upper Chesapeake Medical Center in Bel Air, as the hospital seeks to provide physicians and nurses practice on how they communicate with clients. Effective communication from caregivers is directly tied to how a patient sees their experience, Cancro said, as a better connection can help build a rapport, and make people more likely to adhere to what medical staff recommends.


For Cancro, the product began as she set out to measure the properties of her voice. She previously worked at AT&T Bell Labs, where she got feedback about polishing a presentation that led her to see “the importance of practicing and the importance of focusing on image that you present.” It also grew out of her parenting experience, as she recognized the tone of voice she was using could have a big impact.

“The way that I communicated could really tear them down or build them up,” she said about her kids.

Drawing on a background that also includes working marketing and managing at companies such as Broadcom Corporation and Outsource Laboratories as well as founding Autonomy Engine to develop autonomous control systems for unmanned aircraft, Cancro set out to solve the problem using technology.

The company has a team of seven people working remotely throughout the Baltimore area. Creating the algorithm and platform meant bringing together a team with experience in areas like signals processing, data science, software engineering, machine learning deep learning.

After several years in development, Cancro said the company is now looking to take it to the next level with a focus on sales.

“The idea of giving people feedback about how others perceive them is what makes us really novel,” she said.

Subscribe to our Newsletters
Technically Media
Connect with companies from the community
New call-to-action