Eoin Treacy's view

There was a lot of talk about yesterday and the video ran to almost 40 minutes. That has resulted in a transcript of more than 5000 words which is about double the transcript for the 14th.

Here is a link to the unedited version and the edited version. Sarah has kindly volunteered to edit the transcript. Here is her feedback from doing so today.

That has taken considerably longer to do today. A good few hours.

The problems are:

The software doesn’t always put a full stop or comma in and without punctuation, it’s going to be a difficult read.

It also doesn’t recognise the difference between Rupee and Rupiah, which might be corrected, but if it doesn’t “learn” to recognise it, it could be a confusing read for subscribers!

It doesn’t recognise context so there/their/they’re pretty much all need correcting. Also, we’ve/we’re.

I imagine it’s almost impossible to speak as you do, without a script, and not um and ah, but when you do, the software inserts “at” so you can imagine how many there were to correct.

I am actively trying to educate the system to adapt to my speech but with little success yet. I plan on pursuing this experiment till the end of the week and then we will make a judgement on whether to persist. However, if it takes multiple hours to edit a transcript, that is simply not something we are able to justify.