<time datetime="Error: Invalid time.">Error: Invalid time.
This month's meeting of the FHWA tech group was interesting. I'd not been to one of these meetings before (I did try once, but the meeting was cancelled just after it began). It was a presentation of Graham Wilkie's experience with voice recognition software and how he's been using it to transcribe Perth rates books from the early 20th Century.
The tech group has recently increased it's fees to $5 per meeting, which people mentioned was still less than what most t'otherside genealogical societies charge.
Graham is using Nuance's Dragon Naturally Speaking software (on Windows; $629) to transcribe data from City of Perth rate books from 1907 to 1946. He's been at it a few years, and using voice recognition for a year or so (because of arthritis making it hard to use the keyboard). He uses a web browser and spreadsheet, with the latter overlayed on the former, and he demonstrated the voice commands that he uses to not only enter the text (people and street names, lots of numbers) but also to navigate between cells in the spreadsheet and do a bunch of clever things like adding new rows with prefilled parts. It sounds reasonably easy to create custom commands.
He's done oral histories in the past, with an H1 Zoom recorder, but Dragon isn't good for audio of conversations because each voice has to be set up with its own profile. (I wonder if the two tracks that come out of a Zencastr session would work though, because they are each side of a conversation and are completely separate.)
The rates books were digitized by Ancestry, and although the LGAs (Perth, Vincent, and others I think) have been given the scans they've not got any of the extracted data. Hence Graham volunteering to get it all out and into a structured form. He's a returned data analyst, with an interest in databases. I was super impressed by the SQLite schema that all this text is ending up in! His workflow seems to be to do the transcribing directly into a spreadsheet, and then load that spreadsheet into SQLite where it can be normalized and stored with greater structure and the means of better error detection and other querying. How it'll all ultimately end up being published has not yet been completely figured out, it seems. It'll be published by the SRO though, because the agreement with Ancestry is such that FHWA isn't allowed to host it. (I do wonder about that — surely the images perhaps could be controlled, but the data is a different matter, and in most cases is probably in the public domain.)
The key difference between this data and what Ancestry already provide online is that theirs is all about people, and so is only searchable via names and not by street or individual address. People are very interested in the histories of their houses, and so it's terrific to see that angle being extracted. I want to see if anything similar is available for Fremantle.