Over the last two years, I have had the privilege to speak to the criminal justice classes at San Marcos High School. The teacher tells me that the students look forward to meeting the court reporter more so than any of his other legal professional speakers (court clerk, bailiff, and even judge). When I get to the class, I set up so that they can see my CAT screen on a large monitor, and that allows them to see my steno notes, words-per-minute count, and real-time transcript. I invite the students to write words on the machine and dictate a mock cross-examination while I write.
One of the questions that always comes up is, “Why don’t they just use a tape recorder or something like Siri?” I actually look forward to that question, because the answer is so obvious.
But then I started thinking, what is happening with AI anyway? I know Siri is weak when it comes to accurately writing down the spoken word. She sometimes gets it right, but for the most part there are mistakes, misinterpretation of words, and bad punctuation. So I did a little research.
I came across on article on DZone, “Everything You Need to Know About Voice Recognition Technology.” The article explains how the technology works, and then comments on the “challenges when faced integrating voice capabilities.”
The first challenge mentioned is real-time response behavior. The author talks about network capabilities and real-time response behavior. The network connection, microphone, and server needed to fetch the results have to be optimized.
The second challenge is languages and accents. “Every software doesn’t support all languages, and developers need to identify the regions of their target audience to make strategic decisions regarding languages or accents recognized.” As court reporters, we know how difficult it is for the human brain to decipher what experts are saying with thick accents, and in the world of IP, I find many of the witnesses are from other countries and have heavy accents speaking about the newest technologies.
The third challenge is punctuation. The author writes, “This is one of the biggest challenges that is faced when it comes to voice-based software. Unfortunately, even the best improvements and algorithms may not work because there are virtually endless sentences with different sorts of punctuation.”
After doing my research, I have come to realize once again that the human court reporter who can write clean is essential for making a real-time record. No machine, algorithm, or technology can perform the job of a great real-time court reporter. The students in court reporting school have a bright future with many job opportunities in both the court and freelance sector.
rkramm@veritext.com