Transcripts vs. Closed Captions vs. Automated

For several of my YouTube videos, I noticed that indexing was not working on videos for which I had previously uploaded closed caption files. Initially, my thought was that it may be related to the format in which I uploaded those files (.srt). However, it is now my belief that YouTube was tinkering with this and it is possible that it was turned off/on at some point during the time that I uploaded my closed caption files (in late 2009.)
Since February, I have successfully been able to get indexed for text contained in uploaded closed caption files in any format. In other words, successful indexing was seen for closed captions uploaded in *.SRT (SubRip) and *.SUB (SubViewer) formats, as well as for a plain text transcript uploaded using YouTube’s machine transcription to time-code the text into closed captions. This is worked for older videos where I replaced the existing closed captions with a new file.
As to the question of whether or not Google or YouTube is indexing only closed captions that are turned on by the user or whether or not they are actually using their speech-to-text to automatically index keywords spoken within all Youtube videos, it appears that at this point, only closed captions that are turned on by the YouTube video owner are being used for indexing. I did a test here using a video with a unique phrase of 3 keywords (with no results found) spoken clearly in the video. For this particular video test, I did not upload a transcript or closed captions file. Instead, I checked to see if the video would be indexed for the words that were spoken regardless. Additionally, I did check to make certain that YouTube’s machine transcript was accurately picking up the 3 keywords. The test was unsuccessful and so at this point, it does appear that both YouTube and Google Video search index and factor into rankings the keywords found in caption files but not (yet) the keywords said in videos without captions or transcripts. I.E. – this only works for closed captions that are enabled by the owner.

source: