Podcast: Play in new window | Download (Duration: 8:52 — 4.2MB) | Embed
Subscribe: Google Podcasts | Email | RSS
One of the really great and amazing things about YouTube is the free transcription or closed captioning system they have. It works remarkably well. It uses the same Google Cloud Speech API you can access through other services like Auphonic.com (details about Auphonic’s transcription features).
Warning: If you are uploading video to YouTube you should probably check out what the automatic transcription service is serving up. Most of the time you will be happily surprised, but you may also be shocked when you learn it has a bit of a potty mouth.
Transcripts and closed captions are not identical, but are very similar. Within YouTube the automatic system produces closed captions. For use in show-notes, I prefer a format known as a Production Transcript. If you are uploading to YouTube you can use nearly any text based file that contains time codes and their system will sync the text to the video; you can also upload special CC formatted files such as WebVTT.
Google (via YouTube) now also offers to broker transcription and translation services. For a 1.5 minute video I uploaded for one of my students, (as part of the research for this episode) I paid YouTube $6 for a transcript from Tomedes – Smart Human Translations.es; other offers were priced at $10; there were several offers for me to pick from but they all had a three day turn-around.
In comparison I get a typical 72 hour turn-around from CLK Transcription, Inc., at a price more affordable then those offered by Google. However, it is nice that Google offers both transcription and translation services. CLK charges “$1.02 – $1.50 per minute for non-medical transcriptions.” In this case, the 1.5 min file transcribed by CLK cost $1.75 vs the $6 Tomedes charged.
I ordered the transcript from CLK at 8:32 Pm Sunday and I had the completed transcript 11 hours later. A turn around time that fast is not typical of larger jobs. Here is the file they produced. The file is in the production transcript format (e.g., it’s not a CC file), but when I have saved this format as a txt file and uploaded it to YouTube, YouTube’s in-take system has always done a great job of sync’ing it to my video.
The file from Tomedes was completed at 4:52 Pm or a turn around time of about 21 hours; faster than the promised three-day turn-around. However, it cost 3.4 times more, and took nearly 50% longer that CLK (my regular vendor). Tomedes did not send me a file; they directly installed the transcript as an English CC into my video (an added perk, but probably not worth the extra $5.25). If you view the video, you can see their work as the currently installed English closed captioning.
I was concerned during the process because while I ordered a transcription, their messaging to me referred to it as a translation. Their final message referred to it as English captions. The machine translation that I edited on the 15th (see below), was rendered “un-published” but wasn’t deleted.
Here are some edited screen shots of how I edited the automated transcription and ordered the $6 transcription from Tomedes.

Ordering a transcription for closed captioning within YouTube.

Message indicating that the process to purchase a transcript from one YouTube’s captive partners was completed.

A message that my payment was successful.

Editing the machine translation “keeping queen” to the correct, “keeping clean.”

before fixing the word icebox wrongly spelled as iceboxx.

After fixing the word icebox wrongly spelled as iceboxx.

Example of the user interface for editing a machine generated transcription within YouTube.

The user interface to unpublish or save an installed transcription within YouTube; in this case a machine generated English transcript.

Multiple closed caption files for a single video, one published, the rest, un-published.
I hope everyone has a great week! Bye Bye.
PS. the video I created for my class was made using Adobe Spark.
Leave a Reply