A Conversation with Marcellus Pelcher of VoiceSci

A Conversation with Marcellus Pelcher of VoiceSci

We recently had the opportunity to chat with the man behind Voice Tic Tac Toe, Marcellus Pelcher.

Can you provide a brief bio/background for our readers?

I have a degree in Computer Science. I work at Google as a Software Engineer. I have created six voice applications so far that can be viewed on VoiceSci.com.

What got you interested in the voice space?

Our family had a computer starting when I was very young. I used to love to hear the computer talk. Also, when I saw Star Trek, I wanted to make voice user interfaces shown a reality. When smart speakers came out, I felt science fiction was finally becoming reality. I was compelled to get involved and start developing for them.

You’ve built a number of voice enabled games – are you focused more on the gaming side, or are you looking at other use cases too, and if so which ones?

Yes, currently half of my applications are games. I like games, but I like to do other things too. Some of my other ideas that I want to do are around earthquakes, music note training, astrology, and meditation.

What tips do you have for developers building voice apps?

  • Text and speech are two different things. Don’t treat them as such.

  • Actually read out loud and roleplay with someone your dialog.

  • Design for voice first and then add on visual elements. This will help you to not depend too heavily on visual elements and paint yourself in a corner.

  • Observe what users actually do with your application. Use the interaction transcripts as a source of new features.

  • Talk to your users in real life to get their opinion of your application. While they are using your application, observe things that they do and don’t do.

Where do you see voice apps and devices like Alexa/Google home going in the future?

We are still in 1.0 of what is possible. I see more computation done on the device to reduce the latency of the responses. More control given to developers for things like Text To Speech rendering. More brands using the devices to interact with their customers. People wanting to talk to computers over humans in customer service scenarios.

Connect with Marcellus on Linkedin and follow him on Twitter.  You won’t want to miss him at SuperBot 2018!

Close Menu