From video subtitle to Google snippet

Interpreting a spoken foreign-language interview

  • Apr 28, 2021
  • 0 Comment

I didn't have high hopes that I would be able to understand a spoken three-hour interview that an ancestor had given as Holocaust testimony. Even though my German language skills are improving, interpreting spontaneous speech is much more challenging than written text. The speaker starts and stops, mumbles a phrase, is interrupted, and makes life difficult for the transcriber in all manner of ways.

In this case, lucky for me, the permission-only online stream of the interview from the USC Shoah Foundation, one of more than 54,000 survivor testimonies in its collection, came with subtitles! Not subtitles in English but a written German transcription of the spoken language. By starting and stopping the stream as two or three lines of text appear at a time, I am able to produce a decent translation with the essential aide of GoogleTranslate. 

Besides taking advantage of the software's ability to parse sentences and choose among meanings, I discovered its additional capability to export stored phrases is the key to a very productive workflow for anyone translating a long document. Here's what I did. 

GoogleTranslate is set up to enter short snippets of text at a time, roughly the same length as a screen caption. For best translation, make sure to spell words correctly with the right umlauts. When you have it right, click the star to save the snippet. Then clear the source text and proceed to the next snippet. Continue all the way through a section of the recording. 

Since subsequent snippets are entered more recently, you end up with the items in reverse order when you export the saved snippets to a Google Sheets spreadsheet. To flip them, I numbered the rows and then sorted in descending order. 

From there I could just cut and paste the Translation column of the spreadsheet into a text file, and voila! I have a rough translated transcript of a spoken-voice recording, ready to bring into a word processor for further formatting and editing.