Dexter·Best for procedures, safety and policy steps
Dexter leads with the action and leaves a beat after it, which is what a lockout checklist or a phishing-report procedure needs. Ask him to open a sales kickoff with energy and he comes out polite.
Corporate training and compliance
For a compliance training video I start with my voice Dexter. He stays steady through step-by-step procedures and never rushes the part where a learner has to act. Heisenberg suits longer e-learning voice over, where a module explains why a policy exists, and Derek reads short policy-change bulletins. The real payoff comes at update time. The HIPAA Privacy Rule, for one, requires covered entities to retrain affected staff within a reasonable period after a material policy change, and re-rendering one paragraph is faster than booking a narrator again.
Dexter·Best for procedures, safety and policy steps
Dexter leads with the action and leaves a beat after it, which is what a lockout checklist or a phishing-report procedure needs. Ask him to open a sales kickoff with energy and he comes out polite.
Heisenberg·Best for longer e-learning modules
Heisenberg sits low and gives a concept time to settle, so a twenty-minute module on data handling is easier to sit through. He measured about 175 words a minute in my clips, close to the middle of my voices, but his calm can feel long on a ninety-second microlearning refresher, where Dexter's brisker read fits better.
Derek·Best for policy updates and compliance bulletins
Derek sounds like an announcement, which helps when a short clip has to tell every employee the expense policy changed on the first of the month. He's formal by default, so test names, dates and section numbers before rendering the full set.
E-learning modules
Ten to thirty-minute narrated courses built in an authoring tool and published to the LMS. The narration ships as audio files inside the course package.
Safety and compliance courses
Workplace safety, harassment prevention, data privacy and anti-bribery courses with completion tracking. Clarity counts for more here than personality does.
Policy-change refreshers
Two to five minutes covering only what changed, assigned to the people it affects. This is where synthetic narration saves the most time.
Microlearning clips
Sixty to ninety seconds on one rule or one step, for a weekly email or the LMS home page.
New-hire orientation
Benefits, systems access and workplace rules, updated whenever HR changes a vendor or a form.
Scenario narration for interactive courses
The narrator between choices in a branching scenario. Pair it with questions and knowledge checks, since some states, New York among them, don't count a video on its own as interactive training.
Dexter, about 53 seconds
This script, generated with Dexter on Eleven v4 (0:53)
The narration you'll redo on a schedule.
Not legal advice. Check with counsel for your situation.
None of the rules I checked for this page requires a human narrator. What they do require is content, timing and proof. OSHA wants training in language workers understand, New York wants it interactive, HIPAA wants it documented and repeated after material changes. Your industry may have its own regulator, so check with whoever audits your training before you swap narrators on a required course.
Dexter for procedures and anything with steps, Heisenberg for longer modules that explain concepts. Both are in my calm voices set, and Dexter also leads the customer service voices page if your course doubles as help-desk training.
If you're working to WCAG, yes. Success criterion 1.2.2 asks for captions on all prerecorded audio in synchronized media, at Level A. Section508.gov says auto-captions alone don't meet the minimum for prerecorded media. The good news with synthetic narration is that you already have the exact script, so build the caption file from it. Section508.gov flags speech over about 180 words a minute as possibly too fast for captions, and several of my voices measured above that in my clips, Dexter among them, along with Pulse, Reel, Timber and Bilbop. For captioned training I either pick a voice nearer the middle, like Derek or Heisenberg, or leave pauses between steps, then time the rendered take before I build the caption file.
Keep the script, voice ID, model and settings together with the course source. Edit only the paragraph that changed, re-render it on the same model, and listen to where the new audio meets the old. Update the caption file in the same pass. If you're a HIPAA covered entity, document who took the revised module, since the rule asks for proof of training.
Yes. SCORM, maintained by the ADL Initiative, is a set of standards for packaging e-learning and having it talk to an LMS. A package is a ZIP file with a manifest named imsmanifest.xml at the top level, and audio is one more asset inside it. Export your narration as audio files, place them in your authoring tool, and publish to whichever SCORM version your LMS expects.
In my clips Dexter measured about 190 words a minute, Derek about 180 and Heisenberg about 175, and across all fifteen voices I measured my voices at 178 words per minute. On screen walkthroughs, plan about 20 percent fewer so a learner can watch each click, which puts Dexter around 150. The words per minute calculator turns a module script into a runtime.
Sources