This is not meant to be an ominous, upsetting post but a real look at what is coming. They’ve been saying it for years. Stenographers who have been working for 20, 30, 40 years have said that when they first began working, the warning about digital recorders and artificial intelligent transcription replacing them was a continual threat but something that never transpired into reality. Even ten years ago, this warning didn’t seem real because there was nothing on the market good enough yet for accurate, usable live transcription. Until possibly now.
I was just on my smartphone watching some of my Instagram stories when I noticed an acquaintance had shared his findings of how well the “Live Transcribe” Google app worked. He photocapped the captions of an audiobook he had been listening to that had been transcribed via the app and said, quote: “Most useful app they’ve released since Google Maps”. My curiosity was instantly piqued because I work in the industry of reporting and transcription. Is this app really that great? (And we all know how Google Maps has dramatically changed our lives for the better, so this comparison was especially bold.)
It look me less than a minute to download the free “Live Transcribe” app on my phone and try it out. It was a simple user interface where you spoke directly into the phone, and the captions appeared instantaneously on a black background. There was an option to save the transcription too (for 3 days — why 3 days only?). You could also copy and paste the captions into another document (or e-mail or notepad or whatever).
Here’s the shocking part. The app worked really, really well. I spoke softly, I mumbled a little bit, and tried to trip it up, but it still outputted exactly what I said. I even tried the Cantonese live transcription as well (they have so many languages on there), and it was all transcribed instantly. The transcription, I would say, was about 90% accurate. Um, that’s really, really good. It was all usable text.
I read the reviews on the Google Play store and so many hard-of-hearing and deaf and individuals who need this app for their everyday lives were praising it (with 4.3/5 stars on average). Many commented how it was a lot better than any other transcription app/program they’ve used before, and it effectively allowed them to participate in conversations with others in real time. This is serious. It’s a fantastic invention for the community and those who need it, but what about those of us who do transcription for a living?! Me, other stenographers, and aspiring stenographers?! Is this the end of us?
(One person reviewed, quote: “This is a brilliant piece of software. I used it in school and it’s faster and often more accurate than professional CART stenographers.” Um, this actually doesn’t even make sense and seems more like a personal attack from the person to CART stenographers than a true perspective. Technology will still never be as good as a human stenographer no matter what. Whereas the accuracy for this app might be about 85 to 90%, a human stenographer is certified to caption at a minimum of 98% accuracy. Thus, captions produced by AI are still not better than a human yet. I don’t know what this person is talking about at all. Have they encountered poor work product from a CART stenographer on their off day perhaps?!)
Some stenographers have been saying it for a long time: in order to stay relevant and not be replaced by AI, go realtime. Realtime stenography will not be replaced for a while since AI is definitely still not 100% foolproof just yet and when you are in the legal setting or broadcasting live captions to a giant audience, you can’t afford embarrassing or crass mistakes, the mistakes that only AI would make because a real human would never be confused enough to transcribe “May 10” as “may tent”, for example. If you’re transcribing a Tax Court case, for example, you better know that lawyers will care if the witness said “1110” as opposed to “seven tent”. 😳
Luckily, in the legal world, verbatim output matters 100% more than in casual conversations. In fact, it could be a matter of life or death for individuals on trial if something in the transcript isn’t accurate. This might be one of the last saving grace and essential elements to entrusting a human stenographer to complete the transcripts as opposed to a robot. We real humans are still needed for now. Until the day that artificial transcription is exclusively used across the board for professional settings, they will still need someone to sit there to go through the audio against the text and correct any mistakes in order to deliver a certifiable transcript. That proofreading/scoping job will still exist. Apparently, things in the legal world move slower so maybe in ten years’ time, our jobs will still be secure… but for how long?
What other career can I partake in other than being a stenographer/transcriptionist? Is it time to branch out into something new? How many more years will stenographers be able to continue this very rare but fine and important art?
∴ Ultimately, this app I found is excellent for captioning everyday conversations. I’m assuming it’s likely not effective for captioning technical, accented, or multi-voice settings for professional purposes. However, it just goes to show how well technology is progressing and how stenography might be a thing of the past soon enough. This also means that students requiring CART services will not even need a CART provider anymore. An app like this is so much cheaper, easier to acquire, and faster than a certified professional. 😢😔😓😭😭😭 (The app can only be used when device is connected to the Internet.)
Update: I tried the app again in a room where two people were casually talking on and off (I deleted the transcription afterwards) and the app failed miserably to capture not only verbatim speech but comprehensible dialogue between the two people. The only way this app could work in a conference room is if the mic was attached to each person and they spoke clearly.
Interesting observations: “I could puke” translated into “You’re cute”. XD It doesn’t do well with interruptions. So whereas a court reporter would put dashes to show incomplete speech and thought, the app just continues the transcription or in some cases it backtracks, deletes, changes the previous words into a new sentence. -_- There are missing words. One person went “Uhhh” and the app thought it was a cello playing. XD
I was impressed that proper names and acronyms translated clearly though.