Voice recognition is no longer science fiction. Tell Siri to pull up a song from your favorite artist and it automagically plays (well, about half the time… ). If you have a newer car, there’s a good chance that your auto has voice recognition built-in, so you can send texts, initiate navigation, and much more. Fact is, voice recognition technology has made a great deal of progress, especially over the past decade, and it continues to rapidly improve.
So, what does this technology mean for you, as a transcriptionist? Replacing keyboards and transcriptionists with voice recognition software might seem like something that’s right around the corner. It’s an appealing concept to most people to no longer have to type. Artificial intelligence systems for court rooms are already in use to transcribe testimony at trials and hearings. The next advancement is presumed to include automatic transcribers for telephone and in-person interviews.
However, no matter how advanced these systems become, they still require human intervention. One encounter with Dragon Dictate or other voice recognition programs and you learn that speech subtleties remain difficult for machines. This is especially true with verbatim transcription and the evidential weight placed on it during insurance proceedings.
Verbatim Transcription and Insurance Claims
Claim analysts, subrogation specialists, and special investigation professionals rely heavily on witness and claimant testimony when making decisions. Interviews are an essential tool in this process and in order to interpret it accurately, verbatim transcription is also necessary.
Verbatim transcription does not miss a single syllable—verbal pauses (“um”), false starts, stutters, repeated words, mispronounced words, and grammatical errors are all captured in a verbatim transcript. In most jobs, this is not only unnecessary but a distraction. Interviews for journalism pieces, for example, are never transcribed verbatim because finding usable quotes is time-consuming, if not impossible, if the transcription is verbatim.
However, insurance claims are different. Verbal ticks and pauses can help provide context. Heavy stuttering while discussing one fact could indicate exaggeration or dishonesty. Mispronounced words show a lack of expertise that could reduce credibility for an expert witness. Legal testimony in depositions or court hearings is transcribed verbatim for the same reason—having a complete and accurate typed record provides context and facts that an edited transcript misses.
A Good Tool—Not a Replacement
If you use voice recognition software, you will immediately notice its challenges. The most noticeable ones include:
- Training. This is not just for users who need to learn commands and how to speak in an understandable manner, but the software too. Good quality voice recognition platforms contain an artificial intelligence algorithm that helps them adapt to their users. Particular accents, pronunciation, and idiosyncrasies make the process interesting at first but the software learns to interpret them. There are multi-user platforms available where the voice recognition software saves user profiles so it can apply different rules to each user.
- Names and Specialized Vocabulary: When dictating a report or letter, you stop and give a command to spell names or words that the voice recognition software will not recognize. The adaptive function kicks in here too by remembering new words and names. As you can imagine, there is no practical way to do this in an interview.
- Homonyms and Context: Words may be spelled and pronounced the same but have different meanings. Think “the dog barks” and “the tree has bark” as one example. There are also words with the same pronunciation but different contexts and spelling—take “to, too, two” for example. Voice recognition may understand words but sometimes, its grammar and context skills leave much to be desired.
The key difference that sets verbatim transcription apart from other projects is editing. You can edit that letter, memorandum, or medical record summary after dictating it to your software so that you submit a clean product. When the issues listed above create a head-scratching product, you can fix the errors without changing the meaning.
When you apply these shortcomings to verbatim transcripts, you risk losing important, sometimes crucial, detail. As discussed, verbal pauses and grammatical errors can indicate nervousness or misstating the truth. If your voice recognition software gets verbal pauses or context wrong, and you later try to edit it, your best case scenario is an inaccurate record.
Voice recognition is an excellent feature for devices and particular projects. It makes many tasks easier but verbatim transcription is not one of them. Even the most advanced systems in court rooms require real-time editing by a court reporter. If you are feeling apprehensive about starting a career in verbatim transcription, do not fear. Your abilities to assess subtleties and context will always be superior no matter how much future voice recognition products advance.
I enjoyed your article, but think you are overstating the risk of “grammatical error.” You did not explain the difference between typing and voice writing in terms of the grammatical errors. Certainly when typing those same homonyms can be misinterpreted as well. This is where a careful examination of the work will reveal those issues and they can be corrected either by keyboarding or voice writing. In any event it is up to the keyboarder to figure out which word to use since it is a audio file. From my point there is no difference between keyboarding and voice writing, except in some peoples mind. New technology has always been difficult to accept. Case in point, there are many companies who still demand things be faxed to them so they have an original. Explain the difference between a PDF file printed out and a faxed copy. The email is time stamped and the address is there so again no real difference but just a preference. If you want to stay at the forefront embrace technology and use it, don’t resist it. Sloppy workers exsist in both keyboarding and voice writing. That is where the focus should be. Not on the technology.
Hi David, I’m glad you enjoyed the post and thank you for commenting. I completely agree that technology should be embraced. The intent of the article certainly was not to reject tech. We’re always testing out the latest and greatest voice recognition software and looking for ways to use technology to improve our current workflow.