How time codes work

A timecode (alternatively, time code) is a sequence of times attached to a transcript's word in order to synchronize them to your audio. When our A.I. transcribes your audio sounds into words, it attaches an approximation of a "start" and "end" time to that sounds. For example, the A.I. may think the word "The" was said in the audio between 0 seconds (start time) and 0.1 seconds (end time) with a 98% level of accuracy. 

In this example, you can see how our A.I. synchronized the transcribed word to a time in the audio.

From the above picture to the below, the user changed "jumped!" to "slept." Since the same word was changed, the time codes were perfectly preserved.

Below you can see the user changed "slept." to "was not know to always sleep". The platform had to estimate the time codes, but typically can safely assume the time code is between 1.2 seconds and 1.6 seconds.

You can imagine how much tighter it can get if a whole section is deleted and re-written from scratch. Or, if there are no anchors to attach the previous or next time codes to.

In the worst cases, when time codes are very off, you can adjust the uncertain portions of uncertain text using our Chunk Slider™! Or, have konch run "Re-alignment" and have our AI match your human corrected words to help with the the audio transcription a seconds time around.