In an era dominated by technological disruption, automatic speech recognition (ASR) tools have been widely touted as replacements for human court reporters. Proponents argue that these systems offer faster turnaround times, lower costs, and evolving accuracy. But those on the front lines of courtroom documentation know that in legal proceedings, accuracy isn’t just about words being transcribed phonetically; it’s about capturing meaning in context, intent, and the nuance of human communication.
Nowhere is this clearer than in the following exchange, taken from a real courtroom transcript (Names have been changed to preserve privacy.) involving a recalculation of time and customer data — where words like “cell,” “sale,” and “sales” might all plausibly appear. One misinterpretation by ASR could skew the facts entirely.
ATTORNEY: You’re saying, “balance it out.” But, Mr. Carter, don’t you agree that if you were to redo the —
JUDGE: Just — just redo the calculations on the ones that are highlighted in yellow. You can do that while you’re right there; right?
WITNESS: So there’s 420 minutes in a seven-hour day; right? So a seven-hour day, 420 minutes. Okay? So let’s take a ratio, .8, let’s see, times 7. It’s 5.6 customers per hour, instead of 6.4.
Okay. If you assume, over the course of — pardon me — 5.6 customers over the course of a shift. Okay? Let’s start with that.
So 5.6, times 20 minutes a shift, is 112 minutes. 420, minus 112, 308 minutes of available time, divided by 60, 5.13 hours.
ATTORNEY: And just to be clear, what — the cell that you just recalculated, what does it state in your table?
WITNESS: 5.87.
ATTORNEY: So the number you just calculated is lower than that; correct?
WITNESS: It’s 5.13 hour.
JUDGE: That’s the 20 minutes per customer?
WITNESS: That’s the 20 minutes, yeah.
JUDGE: And that’s for the sales per hour ratio of .8; correct?
WITNESS: That’s correct, your Honor.
ATTORNEY: And so, Mr. Carter, isn’t it true then that all of these cells would go down if you recalculated it with a lower number of hours per shift?
WITNESS: If I used seven, yes. But, again, it goes from 5.87 to 5.13. So from a material perspective, I’m thinking about this and saying, is it really different if it’s 5.8 hours of available time, versus 5.13 hours of available time in the perspective of, can you take a rest break or a meal break? I don’t think it is.
This passage is deceptively simple — but it reveals exactly why trained human court reporters remain essential. In this case, the term “cell” refers to a cell in a spreadsheet. But in a setting heavy with calculations, sales metrics, and labor math, an automatic transcription engine may hear and transcribe it as “sale” or “sell.” It’s a common homophone confusion — one that fundamentally alters meaning.
If ASR makes that substitution, the record is no longer just imprecise — it becomes factually incorrect. Misidentifying “cell” as “sale” changes the nature of the testimony, introduces doubt into the witness’s meaning, and could even mislead a judge or jury about what was actually said. Multiply this by dozens or hundreds of such confusions in a single day, and the credibility of the entire proceeding can be called into question.
The Real Limits of ASR
ASR is a predictive tool, not a comprehension tool. It makes probabilistic guesses about what it hears based on massive datasets. It lacks situational awareness. It doesn’t know that the person speaking is referencing a printed spreadsheet or recalculating math based on a formula. It doesn’t see the pen, the calculator, or the yellow-highlighted cells. It simply hears sounds and makes its best guess.
In contrast, a human reporter sees the witness referencing a document, understands the judicial exchange, and applies logic to the situation. A reporter knows that “cell” refers to a spreadsheet cell because they understand the full context — the topic, the terminology, and the flow of questions.
Comprehension vs. Translation
It’s easy to imagine someone defending ASR by saying, “It’s 90% accurate.” But in legal proceedings, 90% is not enough. Imagine reading a contract that is 90% accurate. Or a medical chart that is 90% correct. In law, every word matters — not just because of its face value, but because of its function in logic, persuasion, and legal interpretation.
This isn’t just an issue of “machine vs. human.” It’s a matter of professional responsibility and legal consequence. Human court reporters are trained not only in stenographic speed, but in ethics, terminology, and situational dynamics. They are officers of the court. Their job is to ensure that the record reflects what actually happened — not just what a machine thought it heard.
Why “Cell” vs. “Sale” Matters So Much
The difference between “cell” and “sale” in the provided excerpt is more than a typo. If the question had referred to a “sale,” it would suggest a financial transaction — something completely unrelated to the act of adjusting shift-length metrics in a spreadsheet. That would mislead anyone reviewing the transcript, especially those not present in the courtroom.
Furthermore, the testimony itself is numerical and layered: it involves minute-by-minute breakdowns, comparisons between 5.87 and 5.13 hours, and labor math involving “sales per hour” ratios. These are not concepts an ASR engine understands — and they are not the type of testimony that lends itself to ambiguous language. This kind of transcript demands a trained human ear and eye.
More Than Just a Typist
The myth persists that court reporters are simply fast typists. In reality, they are guardians of the record. They are trained to understand medical, legal, financial, and technical terminology. They are trained to recognize interruptions, correct speaker attributions, and preserve the integrity of spontaneous speech — all in real time.
In the transcript excerpt, you can see the human interaction: questions, clarifications, and recalculations happening off the cuff. The witness is working out the numbers live. The judge and attorneys are engaging directly. The tone is conversational but carries legal weight. It’s a setting that requires real-time adaptability — something no ASR system can achieve.
The Scopist Factor
I’m a court reporter, and this transcript came directly from one of my own proceedings. I was using a scopist at the time — and even they got these distinctions wrong. My scopist wasn’t a trained court reporter. They hadn’t taken the vocabulary, grammar, or legal coursework that we go through. They didn’t have experience with expert accounting witnesses or SEC hearings, and it showed.
This raises another major problem with relying on ASR and then using scopists and proofreaders to clean up the mess. If those scopists aren’t trained or experienced in our field, they won’t recognize the difference between a “cell” in a spreadsheet and a “sale” in a financial discussion either. The same errors will persist — only with a false sense of confidence that the transcript has been corrected.
In my opinion, the best scopists are former or retired court reporters. They’ve been in the trenches and understand how to preserve the integrity of a transcript. But even then, the reporter must provide oversight. If you’re using a scopist without that depth of experience, you must go over the work with intense scrutiny. We can’t assume someone else caught everything.
Why? Because the reporter is the responsible charge. The scopist and proofreader are subcontractors — they are not licensed officers of the court. They don’t certify the record. We do. Every word of that transcript ultimately comes back to us. We review it, correct it, and sign our name to it. And very often, we catch dozens of errors — some critical — that others miss. We don’t always report those corrections. Sometimes we notify the scopist. Sometimes we let it slide. Sometimes we fire them. But the bottom line is: we are the ones who are responsible for the accuracy of the record.
If you think the only work a court reporter does is hand off audio to someone else to transcribe and polish — you’re dead wrong. We are the last line of defense for accuracy and truth.
The Bottom Line
ASR can be helpful in many domains — meetings, lectures, casual interviews. But in a courtroom, where truth must be preserved without error, we must not be seduced by convenience. A machine might get the gist. But when it comes to justice, the gist is not enough.
One wrong word — like “sale” instead of “cell” — can alter the facts.
This is why court reporters matter.
This is why accuracy is human.
And this is why, in the balance between cost and credibility, we must continue to trust those whose job it is to listen carefully, think critically, and transcribe faithfully.
If you have transcript portions or real-world scenarios where ASR would likely fail — homophones, technical language, overlapping speech, or nuanced phrasing — I’d love to see them. Send them in. Let’s collect these examples, write about them, and show the world why human court reporters remain irreplaceable.
Disclaimer
The content of this post is intended for informational and discussion purposes only. All opinions expressed herein are those of the author and are based on publicly available information, industry standards, and good-faith concerns about nonprofit governance and professional ethics. No part of this article is intended to defame, accuse, or misrepresent any individual or organization. Readers are encouraged to verify facts independently and to engage constructively in dialogue about leadership, transparency, and accountability in the court reporting profession.
- The content on this blog represents the personal opinions, observations, and commentary of the author. It is intended for editorial and journalistic purposes and is protected under the First Amendment of the United States Constitution.
- Nothing here constitutes legal advice. Readers are encouraged to review the facts and form independent conclusions.
***To unsubscribe, just smash that UNSUBSCRIBE button below — yes, the one that’s universally glued to the bottom of every newsletter ever created. It’s basically the “Exit” sign of the email world. You can’t miss it. It looks like this (brace yourself for the excitement):
STENOIMPERIUM 