On May 1, 2016, at 4:21 PM, Levy, Michael <[email protected]> wrote: > > Peter, if you are interested in speech-to-text software, let the list know > and I'm sure there are people who can share their experiences.
Automated speech to text can’t yet do better than about 80% accuracy under ideal conditions. Most recordings are not ideal. 80% accuracy is often cited as the minimum to be consider usable. Greater than 80% requires some form of “training” or human moderation. One trick used in the interview business is to have the subject read the standard training “script” at the beginning of the interview. This is used to train the software to each voice prior to transcription. Obviously this requires planning and is not particularly convenient. Nuance: Makers of “Dragon Naturally Speaking” is arguably the market leader in voice-to-text. Google/YouTube incorporates automated transcription and can be used as a “free” solution. It’s a bit cumbersome use this method at large scale. Commercial, web based transcription services have exploded. These services have made professional transcription much cheaper. This company markets itself as a captioning service, but you can use the service for simple transcription. I have not yet used them, but I do like their system. https://www.syncwords.com/ Cheers, tod _______________________________________________ You are currently subscribed to mcn-l, the listserv of the Museum Computer Network (http://www.mcn.edu) To post to this list, send messages to: [email protected] To unsubscribe or change mcn-l delivery options visit: http://mcn.edu/mailman/listinfo/mcn-l The MCN-L archives can be found at: http://www.mail-archive.com/[email protected]/
