Andy Plesser of beet.tv has the story on Podzinger's use of its speech-to-text technology to make 1.5 million YouTube videos less opaque to search engines. According to Andy's post, Podzinger (a subsidiary of BBN, a major defense department contractor that has used the same technology to analyze audio for three letter government agencies) analyzes the YouTube clips to determine if they have music or speech.
If speech is detected, a transcription is created with close to 80 percent accuracy, according to the company, and Podzinger organizes them, using the embedded YouTube code, on its public portal with the transcripts. The company is selling contextual ads around the YouTube embedded content.
Andy has a beet.tv video conversation with Podzinger CEO Alex Laats.