Google AI Helping With Transcription

I have been transcribing for BTP for 7 1/2 years and recently started using Google AI for help in figuring out some words. If you have part of a word and ask it a question it may help you come up with the word. Many times I have to ask several questions but if you are patient and persistent, it may help. I have also had luck with French words where I post the whole phrase and then tell it I am transcribing this phrase and need help with the word that ends in moins for example. It has also helped with names and places. But, as I said you may have to ask several questions.

1 Like

Great idea! Thanks, Henry!

I have been using French spell check in a similar manner to review documents that have been mis-transcribed. I can read French pretty well, so often I can parse the correct word even if obscured, but spell-check can offer many similar options to choose from - and a bonus for always giving the correct accent marks!

2 Likes

How do you get to French spell check? Is there a setting somewhere or do you have a special program?

I usually set up a blank Pages document, then set region/language to French/France (you can do the same in MS Word), then copy in document and choose “edit/spelling and grammar/show spelling” and go through checking things.

Of course you still need to review each line by eye after that, but I find this saves a LOT of time, especially with entries that ignore all accent marks!

2 Likes

Thanks Diane. That’s a great tip!

Yes! The hard part, though, is to resist correcting the text in case it’s misspelled. Still … it IS helpful to know what these names really are.

1 Like

If can read the misspelled name, as you know, transcribe it as written but I agree that knowing the real name is helpful. For me, if AI gives me a potential name, I try to see if it fits what I am seeing in the transcription and then Google the name it gave me. Sometimes AI is wrong so I need to be prepared to dive deeper.

1 Like

Yes, indeed - I cross check all the spelling suggestions before accepting them to see if they are accurate to the text. Still a massive time saver, even so.

1 Like