Why does closed captioning goof up?

A: Here at SnapStream HQ, we are deeply entrenched in closed captioning since it’s a key ingredient in our TV search juice. But every so often, we’ll see mistakes in the captions, from a generic typo to a full-blown word massacre. You know, the ones that end up as CC Fails for their inappropriate nature. So is it human or machine error?

It largely comes down to the way the captions were made. 

Real-time captioning. While a faulty decoder or poor signal can produce captioning errors, more often than not they are the result of human error, particularly during live programming. Stenographers are the turbo typists who churn out captions faster than 300 words per minute on a steno machine, which contains 22 keys and code-based phonetics for every word. During live broadcasts, such as the local news and sporting events, every keystroke counts, as all it takes is a single incorrect keystroke to type the phonetic codes for two completely different words. Homonyms and unusual names can prove especially difficult.

Prerecorded captioning. It can take up to 16 hours to caption a one-hour prerecorded program, such as a movie or it sitcom, because captioners use special software to set the placement and duration of the caption on the screen. Thus, you tend to see minimal mistakes.

Speech to text. In smaller markets, TV stations may commonly use non-human transcription methods, like speech-to-text and speech recognition. Though technology has improved significantly, anything automated can’t accurately capture all the nuances of language without some degree of error.

SnapSream shall overcome. In most cases, SnapStream’s TV search still manages to be very effective with boolean logic. For instance, you can perform perform a “fuzzy” search using the syntax operator ~ to generate results for similar sounding and misspelled terms in the closed captioning data. For example, SNAP~ returns snap, soap, swap, snag and slap. That’s the power of our TV search technology, which blends not only closed captioning information, but TV listings information as well.