Monthly Archives: March 2016

Will Text to Speech Destroy Accents

Audio to text technology has finally reached a point where it doesn’t require a person to repeat themselves a half dozen times. I’ve worked on some of the technology to help improve this specific quality. I’d listen to hours of transcripts and transcribe them, matching them up with what the software would spit out at me. The primary issue that this technology is suffering through now is acknowledging accents. For example; I help review the results of the software’s transcription. I hear the audio request made by the user but they have such a thick South Texas accent that even I am not able to understand exactly what they’re trying to say. Usually, I would mark ‘accent’ and the system at work would ask me to identify the accent if I could. Continue reading