Major errors in speech recognition system

A study of 615 breast imaging reports has revealed more errors in those prepared using a speech recognition tool, than using traditional dictation. The reports which were produced with the aid of an automatic speech recognition tool had eight times more errors than the reports which were dictated to a transcriber.

The reports were studied between January 2009 and April 2010, using reports generated from Toronto’s University Health Network. They were split into two sections, the first half were produced using the traditional method of a radiologist dictating the report, which is later transcribed by a transcription team. The other half of the reports were produced using automatic speech recognition, which instantly transmits the dictation from the radiologist onto a computer screen.

The main author of the study group, Anabel Scaranelo, stated that the results of the study revealed a requirement for time spent proofreading and editing. The study proves that, although the speech recognition tool saves time, more time needs to spent on the proof reading. At least one mistake was discovered in almost a quarter of the reports produced with ASR, while only one in four percent were found in the reports produced by conventional methods.

Business owners who are struggling for time may be tempted to use methods which produce copy in a shorter period of time. However, as the study shows, the number of errors in the copy will create a poor image for your business. Content outsourcing to professional SEO writers will ensure accurate copy which is original and has been thoroughly edited.

Leave a Reply