Speech for Super Tuesday

On February 5, the New York-based public radio station WNYC will have extensive coverage on Super Tuesday. To optimize listener interactivity, the station will leverage SpinVox’s speech-to-text (STT) technology during an hourly segment called "America’s Exit Poll." 

During the segment, voters can call into a special phone number set up for the radio station. SpinVox will convert spoken comments into text and provide the hosts, John Hockenberry and Adaora Udoji, with that text. Hockenberry and Udoji then have the option of reading the messages on air.

"By using (SpinVox’s) voice-to-text system, every listener call will become part of the conversation," John Keefe, WNYC’s senior executive producer of news, said in a statement. "Not just the few that typically get through during call-in shows."

According to SpinVox spokesperson Tony Carter, the company engineered the infrastructure to handle 60 calls simultaneously—a far more efficient solution than having a bank of people trying to handle phone assignments.      

While a common complaint with STT technology is that the voice recognition component isn’t reliable enough to provide a completely accurate transcription of the conversation, Carter says that’s a non-issue for the service SpinVox provides. 

"The idea has never been to create a transcription service but instead a hosted message service that provides a high level of accuracy to allow readers to understand the message," he says.  The speech recognitions are also bolstered by humans who listen to bits of the message in case of ambiguity. Thus SpinVox’s STT is accurate, but also provides enough automation for large-scale events. 

Listeners can offer comments and stream live coverage at www.americasexitpoll.org.