How to make voice, accessibility, and AI decisions that hold up under pressure
In my last post, I talked about the “Friday-night problem.”
The scramble. The comment pile. The victim-y heroics.
And the way “we have to” turns choice into burden, which turns the work into minimum viable effort.
Here’s the hinge that changes everything.
Most teams need shared rules.
The simplest “rules” I know are two words:
Context And Intent
Context is what happens right before, and right after.
Intent is what the moment is trying to make the audience feel and understand.
That’s it.
No jargon. No long meetings. Just a shared way to decide.
A simple example:
Someone slaps someone else.
That slap can mean ten different things, depending on context.
If right before, they were playfully dancing around, the slap might read as teasing. Or flirtation. Or a joke that went too far.
If right before, two enemies faced off in a high-stakes standoff, the slap can read as domination. Or betrayal. Or escalation.
Same action. Different meaning.
That is context.
Now add intent.
What is this moment trying to do to the viewer?
Make them laugh?
Make them flinch?
Make them worry?
Make them understand who has power now?
When a team is aligned on context and intent, the “voice problem” stops being subjective.
Notes stop being taste wars.
Instead of “I don’t like it,” it becomes, “Does this match what the moment is doing?”
That shift is agency.
It is also the secret sauce of performance.
I’ve watched context and intent light up voice actors, executive rooms, training sessions, and speaking engagements. Not because it’s complicated. Because it’s actionable. You can use it today.
Two Questions I Use In Real Conversations
If you want the tool without the philosophy, here it is:
1. What do we want the audience to feel and understand in this moment?
2. What are we doing that pulls attention away from that?
Those two questions clean up a surprising amount of noise, including:
Why audio description sounds flat or rushed even when the script is solid
Why a voice choice feels “fine” but not trustworthy
Why a synthetic voice works in one place and breaks the spell in another
Why a team keeps re-litigating the same decision every release
The all-or-nothing trap with AI voices.
I’m not interested in schooling the industry. I’m in it. I understand why teams cling to all-or-nothing thinking.
Binary decisions feel safe.
Always use AI.
Never use AI.
It gives people something concrete to hold.
But it also replaces thinking with a stance. And stance does not protect trust.
The decision is rarely about the technology alone.
It’s about the moment.
Sometimes a synthetic voice is perfectly fine, especially when the job is informational and the emotional load is low.
Sometimes it’s a terrible fit, especially when the moment needs warmth, grief, humor, tension, intimacy, or precise human rhythm.
Most teams only debate cost and speed. That’s understandable. Those are the numbers in front of them.
But “cheap and fast” is not the same as “lands well.”
Also, if you are adopting synthetic voices without a clear plan for quality, oversight, and accountability, you are making a brand trust choice.
Why Audio Description Is The Fastest Truth Test I Know
Audio description forces the real question:
What matters right now, and how do we make that land?
You cannot hide from that question.
If a team is unclear on what matters, the audio description will expose it immediately. Pacing gets weird. Tone gets inconsistent. Review becomes subjective. Notes become endless. (And on many teams, that’s exactly when accessibility gets treated as “extra” instead of “part of the storytelling.”)
If a team is clear on what matters, audio description becomes an accelerator. It pulls the whole workflow into focus.
This is one reason accessibility standards even talk about audio description for video, because it directly affects whether content is perceivable for people who cannot see the visuals.
A Few Questions People Ask
When Should We Use Human Voice Vs. AI Voice?
Start with intent. If the moment needs emotional nuance, human rhythm, or trust through tone, default human. If it’s low-emotion, informational, and you can maintain consistency, AI can be fine.
Why Do Vendors “Hit Spec” But Still Miss The Moment?
Because specs can be checked, but intent has to be understood. If your notes do not name intent, you get technically correct work that feels off.
What’s The Smallest Change That Reduces Risk Fast?
Agree on context and intent before notes go out. One shared sentence can save ten rounds of feedback.
Why I Built The Sprint
I’ve always enjoyed these conversations. I’m generous by nature. I like solving the puzzle with people.
Now it’s cleaner.
It’s a defined process. Clear steps. Clear outputs. Clear next actions.
Not more meetings. Less scrambling.
A simple way to stop bolting things on at the end.
A way to move from “we have to” to “we get to.”
If this is familiar, and you want the 30-day reset that turns scattered opinions into shared rules, the Sprint page is here: roysamuelson.com/sprint.
One question to sit with:
Where are you relying on being the hero, instead of shared rules?