In a recent deposition, a witness gave what seemed like a minor detail in their testimony — but that detail reveals a significant flaw in the use of automatic speech recognition (ASR) systems in legal settings. The witness said:
A: “Well, in theory, you know, he’s just flagging shifts, but if, for example, you flag a shift that says, okay, this is at five hours and two minutes, if the opportunity was at four hours into the shift, and they chose not go then, and they chose not to go at four hours and 15 minutes, four hours and 30 minutes, four hours and 45 minutes, and they went at five oh two, well, it’s a different context under looking at whether or not the opportunity was provided.”
At first glance, this seems like just another rambling sentence in a long day of testimony. But take a closer look at how the witness phrased the critical time marker: “five oh two.” They did not say “five hours and two minutes” directly the second time. They didn’t say “5:02.” They said it casually, imprecisely, in natural speech — the way people often do in everyday conversation.
A human stenographer understands this. A certified reporter, trained in real-time transcription, hears “five oh two” in context — surrounded by references to four hours, four fifteen, four thirty — and knows intuitively that the speaker is referring to five hours and two minutes into the shift, not 5:02 p.m. or a case number 502 or section 502 of a penal code.
But an ASR engine?
More likely than not, it would transcribe “five oh two” as “5:02” — a time of day. That’s not just a small formatting difference. That’s a potential misrepresentation of fact. If a legal argument hinges on when a decision or action took place during a shift, that subtle difference could change the interpretation of intent, timing, or negligence.
And here lies the deeper issue: ASR has no idea what the speaker meant.
The Illusion of Accuracy
Automatic speech recognition relies on massive language models that predict likely words and phrases based on sound patterns and statistical frequency. It doesn’t understand language. It doesn’t grasp context. It doesn’t know the difference between “five-oh-two” meaning a timestamp, a duration, a quantity, or an arbitrary number.
And yet, to the untrained eye, the transcript it produces looks perfect. It’s grammatically clean. It’s punctuated. There are no spelling errors. It creates the illusion of accuracy. But in court, illusion isn’t good enough. The record must be precise, contextual, and verifiably accurate. There is no room for interpretation errors masquerading as clarity.
This is why court reporters — real humans — remain critical in the courtroom and deposition suite.
Stenographic Intelligence vs. Artificial Guesswork
When a trained court reporter hears “five oh two,” they aren’t just hearing sounds. They’re interpreting meaning based on the surrounding discussion, the pace of questioning, the subject matter of the case, and even the tone of the witness. They know if a case involves shift work. They know if the line of questioning is about timing. They don’t just take down words — they take down meaning, word for word, as spoken, in context.
Compare that to ASR, which hears “five oh two” and might decide to write:
- “5:02”
- “502”
- “five or two”
- “five O two”
- “five-oh-two”
None of these are inherently wrong in other contexts. But only one of them is right in this context — and ASR is guessing.
Worse yet, once that transcript is turned over as official, the error is baked into the legal record unless a human proofreader — usually at extra cost and time — combs through and identifies the mistake. Even then, what happens if the human doesn’t catch it? Or what if there’s no recording to verify?
What happens when a case outcome depends on that timing detail?
The Legal Stakes of “Five-Oh-Two”
Let’s assume this scenario plays out in a wrongful termination case involving labor violations. An employee was supposed to be offered a break or medical attention four hours into a shift. The employer claims the opportunity was provided, just not accepted. The timeline matters. If the employer didn’t offer that opportunity until five hours and two minutes into the shift, that’s potentially a breach.
If the ASR transcript says “5:02,” a defense lawyer might argue that it simply reflects an evening timeline, perhaps during a night shift. But if a court reporter correctly captures “five-oh-two” as referencing five hours and two minutes, in context with four-fifteen, four-thirty, and four-forty-five, that paints a different picture entirely.
A single transcription choice could sway the case.
That is the real-world risk of replacing trained reporters with automation.
Verbatim is Not Enough
ASR fans often argue that the technology is “good enough” and “always improving.” They point to better error rates in ideal conditions. But court isn’t ideal. Court is full of accents, background noise, people talking over one another, legal terminology, emotional outbursts, and, most of all, real-life ambiguity.
“Good enough” isn’t the standard. “Verbatim” isn’t even the standard. The standard is accuracy, context, accountability, and reliability. That can only come from a human professional who knows what’s at stake.
Court reporters don’t just write what people say — they preserve what they meant, precisely, under oath, on the record. They make judgment calls, distinguish dialects, clean up mumbling, and even ask for clarifications. ASR does none of that. It can’t. And until it can, it doesn’t belong in charge of the record.
The Cost of Convenience
The push to replace reporters with machines is usually driven by cost — not quality. But how many cases can we afford to get wrong before the costs of those mistakes outweigh the savings?
If you’re a lawyer, litigant, or judge reading this: Don’t accept convenience over clarity. If a case mattered enough to file, it matters enough to record right. And if you’re a court reporter, don’t stop advocating. The next time someone says your job could be done by a computer, just say:
“Really? Ask it what ‘five oh two’ means.”
And see how many answers you get.
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 
Hi there, I just started receiving your emails a couple weeks ago. Not sure how I got signed up or how that started, but I have read a few of your emails and I’ve found them somewhat interesting. I’m just curious, who is this, who is the person behind all these emails, and what is your purpose, mission statement, goal, etc.? Thanks, April
LikeLike
Hi April,
Thanks so much for your message—and for reading the emails. I appreciate your openness and curiosity.
You’re receiving these emails because, at some point in the past, you connected with me through a professional effort or industry-related outreach where you opted in to receive updates. I’ve continued that mission through StenoImperium—a movement focused on strengthening and protecting our profession for the long haul.
This movement exists to help court reporters reclaim our identity—not just as service providers, but as the Responsible Charge of the record. We’re aiming to shift how we’re seen in the legal system: not just as support, but as essential partners in ensuring justice.
I’m grateful you’re here and hope you’ll continue on this path with us.
Warmly,
StenoImperium
LikeLike
More accurately, ASR is a court reporting tool ~ it’s not the be-all-end-all. Obviously, it’s not your business model, but it’s an extremely successful (efficient and intelligent) business model now in 2025 and beyond. ASR dispenses with the machine only; it doesn’t dispense with full-audio editing and proofreading afterwards — as you know, the transcription process consists of three steps: capture, edit, proofread, for accuracy! AI is “artificial intelligence,” and it’s not infallible — it’s not “accurate intelligence”! ASR and AI are distinct, btw – and I believe you are perpetuating fallacies all around on ASR and AI, on a continuing basis. Please do more fulsome research on ASR and AI and their distinctions, especially concerning court reporting, because objectively there are glaring inaccuracies in your current findings.
LikeLike
Thank you for your thoughtful comment and for engaging with the article.
I agree with your point that ASR (automatic speech recognition) is a tool—not inherently good or bad, but highly dependent on its application, limitations, and oversight. However, I must respectfully challenge the assertion that it’s merely the machine being replaced. In many jurisdictions, ASR is marketed and implemented as a full-service court reporting solution, not just a capture mechanism. In those cases, there is no certified reporter or Responsible Charge overseeing the process. That’s where the danger lies—not in the tool itself, but in the removal of human accountability.
You’re absolutely right that transcription involves three steps: capture, edit, and proofread. That model works when qualified humans are involved in all three phases. But when corporations rely on bulk ASR output with minimal human correction, often outsourced or rushed due to volume, the integrity of the record suffers. As we’ve seen from transcripts like “five oh two,” context errors may appear minor—but in litigation, context is everything.
As for distinguishing ASR from AI, I appreciate the clarification. Yes, ASR is one branch of AI, and it’s important to avoid conflating terms. That said, my criticism is directed more at how these tools are currently being deployed in high-stakes legal environments without adequate checks and balances—not the technology itself in theory.
I always welcome pushback when it’s meant to elevate the conversation, and I invite further discussion around how any technology in this space can be used responsibly, ethically, and accurately—especially in our courts of law, where people’s lives and liberty are at stake.
Warm Regards,
StenoImperium
LikeLike
Well, I would never believe ASR is a complete translation solution ~ it’s too fallible. I use RSR as my “machine,” but am treated as though I’m using it as the “capture/edit/proofread” package. I sweat over my transcripts, and don’t mind them being held up to scrutiny. This is problematic for those of us advancing with the modern era — so odd! — ASR is a tool. Ohhhhhh, I’m so sorry — literacy isn’t on the rise…! No one knows better than that ASR is incomplete? ! ? Wow — however, don’t beat me up, but I’m over steno! There’s a new kid in town.
LikeLike
Chris, Final: RSR isn’t designed to be used in the method you propose.
LikeLike
Thank you for your comment and for raising that point.
To clarify, I’m well aware that RSR—real-time speech recognition—is not designed to function without qualified human oversight. However, the concern I raise is not about the intended use of RSR but rather the current and increasing trend of how it’s being deployed—particularly in legal settings where it’s marketed as a cost-effective replacement for human court reporters.
The issue is not the design or the technology itself, but the implementation without the checks and balances of a certified stenographer or trained transcript editor. The error in the transcript example I referenced (e.g., “five oh two”) underscores this point. It’s a real-world consequence of removing professionals from the process or minimizing their role to post-hoc cleanup, often under tight deadlines and without firsthand context.
RSR may be efficient when used appropriately—with human review, context, and accountability—but in practice, that’s not always happening. That gap between design and deployment is precisely what I’m calling attention to.
Appreciate your perspective, and I’m always open to dialogue about how we can improve the integrity of the legal record—whether through technology, training, or truth-telling.
Best,
StenoImperium (not to be confused with Christopher Day and his blog Stenonymous)
LikeLike
The court reporting industry needs to promote the new technology and recruit English majors/college grads for the workforce. Talent can look elsewhere for a career. Why not make court reporter a career worth having w/intelligent and efficient software: RSR?
LikeLike
I had a plaintiff say he had been convicted of burglary too, or did he mean burglary two? I had to ask for clarification. ASR wouldn’t have caught that.
LikeLike
Chris, You misquoted –and incorrectly framed — my last comment/reply to you in your most recent writing. For clarity, I would like you to publish my last comment/reply to you, verbatim, here in this article.
RSR is Realtime Speech Recognition, the next iteration of the Eclipse technology/software I’ve been using since 1987.
Question: Do stenographers (really) want to work their lives away?
Lastly, please do due diligence concerning RSR. Thank you ~
LikeLike
Thank you for your comment and for providing additional clarity on your use of RSR—Realtime Speech Recognition—as an extension of the Eclipse platform you’ve worked with since 1987. I respect your long-standing experience in the industry, and I appreciate your engagement with this conversation.
I will consider publishing your original comment in full, as I value transparency and diverse perspectives. However, I also maintain editorial discretion when it comes to which comments are included in article bodies versus comment sections.
As for your question—“Do stenographers (really) want to work their lives away?”—I think it’s important to reframe that. Stenographers, like many professionals who take pride in their craft, want to protect the integrity of the legal record and ensure due process is served. The work isn’t just about longevity or labor; it’s about public trust, accuracy, and accountability—especially in environments where people’s lives and liberty are at stake.
That’s why due diligence around any emerging technology—including RSR—is critical. I welcome innovation. But innovation without oversight or ethical guardrails is not progress—it’s a risk. And when RSR is implemented without a trained, certified professional in the loop, the risk to justice and transcript integrity becomes real, as we’ve seen in real-life transcript failures.
I appreciate your encouragement to continue researching RSR. My work is driven by ongoing inquiry, not static assumptions—and I invite continued discourse on how we can merge experience, ethics, and evolution in our field.
Respectfully,
StenoImperium (Not Chris, or Christopher Day, the blogger of Stenonymous)
LikeLike
Thank you for the suggestion to reframe my question — you are correct. I can’t figure out why the pushback to advancing CR technology, especially towards me personally. Remember: 7% of steno students graduate steno school….
I recommend embracing the latest intelligent and efficient technology and recruiting English majors and highly-literate talent into court reporting ~ because this field has to compete for an educated workforce at the end of the day.
LikeLike
Thank you again for your continued engagement and thoughtful response.
You raise a valid and important point about graduation rates—yes, the 7% statistic is troubling, and it speaks to the need for reform in both training and recruitment. I’m in full agreement that we should be attracting literate, articulate, and detail-oriented individuals to this profession. Where we may differ is in how we preserve the integrity of the legal record while adapting to the future.
Advancing technology isn’t inherently bad—in fact, I welcome tools like RSR when used alongside a trained, credentialed human professional who is ultimately responsible for the record. The resistance you perceive isn’t about clinging to the past—it’s about protecting against the erosion of quality, context, and due process when humans are taken out of the equation entirely.
Court reporting is not just about efficiency—it’s about accuracy, judgment, and trustworthiness in high-stakes environments. We don’t oppose innovation—we oppose substitution of human skill with unchecked automation.
Let’s work together toward a future where the profession does attract top talent—college grads, English majors, and others—by promoting the value of our work and the ethics of human-led transcription, enhanced by technology, not erased by it.
Warm regards,
StenoImperium
LikeLike
I’m def a collaborator ~
LikeLike
Thank you—and I truly appreciate hearing that you’re a collaborator. It sounds like we’re very much on the same side: the work of preserving the record should be done by professionals. Where we agree wholeheartedly is in advocating for intelligent, highly literate individuals to join our field—and for those professionals to be empowered with the tools they need to succeed.
Court reporters should absolutely have access to innovative tools—whether it’s ASR, AI-driven research, recording devices, or even platforms like ChatGPT—to enhance our productivity, accuracy, and service to the courts. After all, doctors aren’t replaced by surgical robots—they use them. So why should we be treated any differently?
The future of court reporting must include open, honest, and respectful dialogue. Suppressing ideas, shutting people down, or ostracizing those who raise difficult questions doesn’t make us stronger—it holds us back. If we truly want to compete for top-tier talent and ensure the integrity of the record, we need to create a profession that welcomes both tradition and innovation, experience and evolution.
Let’s keep this conversation going. Our profession needs more collaborators like you.
With respect,
StenoImperium
LikeLike
Can you send me a link to reset my password so I can see the password-only messages?
>
LikeLike
Chris, My comments are attributed to dreamilyabe-something —- I’m Susan Ashe!
LikeLike
I’m not Christopher Day. He writes the blog called Stenonymous.
LikeLike