Speech Recognition Technology: Past, Present, and Future
Transcription began with doctors dictating into recorders and transcriptionists using Dictaphone machines to type out the report using a typewriter on carbon paper. Mistakes were difficult to correct. Then it evolved into electronic sound files sent and electronic hand-typed reports being emailed back. Now technology has advanced so much that software can literally type out what a doctor is saying and achieve about 85% accuracy after learning the doctor’s speech patterns and habits. Learning software! What’s next? An artificial intelligence named HAL mechanically intones “Good morning, Dave” to every doctor and types out what you’re going to say before you even say it?
Speech recognition is very much in use by today’s physicians. What this means for the growing body of qualified medical transcriptionists is that in order to stay competitive in this ever-changing field you might consider diversifying your skill set. Some transcriptionists will need to transfer their skills to editing medical reports rather than transcribing them. In 2009 Career Step conducted a survey of 187 companies and what we discovered is that 30% of them use SRT (speech recognition technology). More and more MTSOs are having their MTs do a combination of straight transcription and editing. M*Modal, for example, have their MTs do, on average, about 40% straight transcription and 60% editing.
Believe it or not, the concept of speech recognition technology has been around since the 1930s! At the 1964 World’s Fair in New York, IBM unveiled a speech recognition program called “Shoebox” that could respond to 16 spoken words, converting them to electrical impulses. It was a remarkable beginning. Eventually the military recognized the benefits of this technology and unloaded some funding into researching its capabilities. By the end of the 1970s, SRT was available for commercial use for a pretty penny. Installation and use of the technology could cost up to $100,000!
From here, SRT evolved into two different methodologies: automated messaging responses tailored to call center use and speech-to-text applications. More and more companies were using automated systems to troubleshoot calls or field them in specific directions in order to better utilize human representatives. The speech-to-text possibilities were slower in coming to light. Dragon Systems, now an industry leader in the technology, was one of the first to develop a PC application capable of recognizing 8,000 words.
Today, speech recognition capabilities are being explored and expanded in many different industries and venues. Not only can your computer and phone talk back to you, now so can your car with the advent of systems like Synch. The possibilities are seemingly endless and new ideas are being generated every day. As the technology improves, so does the price. The viability of using this software as a production enhancement has become a reality for many physicians. Recently Career Step introduced the new Medical Transcription Editor program, which utilizes SRT software developed by one of the industry leaders, M*Modal. The Medical Transcription Editor program was created to answer the growing demand of transcription companies for MTs to have editing skills added to their MT repertoire.
Ultimately, the thing to remember is that SRT does not replace MT. MTSOs, demand an accuracy rate in the high 90th percentile—far above the best any VR system can produce. Maybe at some point down the line if humans are reduced to batteries or robots are running around as personal servants, then—maybe then—a form of speech recognition technology may be a threat to human medical transcriptionists. Until that date however, nothing will replace the human brain, so rest easy!