Skip to content
voicefieldguide

AI voices for a cyber incident customer update

Updated November 6, 2026

The picks

Press play to hear each voice read this guide's sample line.

These picks compare my fifteen voices. I own them, and ElevenLabs may pay me when you use them.

Derek·Top pick for a short bulletin

Firm broadcast pacing gives a concise status read a clear shape. Derek's formal register can sound like a press release on a conversational service page.

0:09

Dexter·For customer action steps

Dexter's steady support delivery leaves room for an instruction and the route that follows. His delivery stays polite when a line needs sales energy.

0:10

Heisenberg·For a longer explanation

Heisenberg's low, even read gives a longer explanation room to land. The space after each line can overrun a brief update slot.

0:10
Add Derek to My Voices

Derek is my first pick for a short cyber incident customer update because his firm broadcast pacing makes a concise bulletin easy to follow. Dexter suits a line of customer actions, while Heisenberg has room for a longer explanation. The human incident lead still approves every fact. Keep a named human spokesperson available for questions.

What work should the voice carry while facts change?

Keep the job narrow. A short customer update can state the confirmed service impact, what the business is checking, the next update time and where customers can find the current version. When a relevant detail remains unknown, label it unconfirmed and say when the next review is due if the incident lead has approved that time. The voice cannot make an uncertain statement safer.

The date attached to a news story can differ from the date of the event it describes. On October 20, 2026, the ICO published a news item about reprimands issued to Ogi Networks and Data Cymru. The underlying ransomware attack occurred on November 1, 2024. Read the ICO item.

This page covers a communications producer’s casting, approvals, revisions and playback checks. It gives no security response instructions or notification deadlines. The corporate training guide covers employee procedures. A customer incident update needs its own approval and version trail.

Which of my voices fits each update?

Derek for a brief bulletin

Derek has a straight news read and firm pacing. I would start with him when the update needs a clear opening, one status and a next-update time. Plan near 180 words per minute, his measured rate in my clips, then leave room for the actual service name and a deliberate pause before the time. His formal delivery can feel like a press release on a casual support page, so keep the copy plain.

Dexter for a customer action

Dexter is steady enough for step-by-step instructions. He fits a short line that tells customers where to check status or how to reach the usual support team. His delivery stays polite when a line needs sales energy. Keep each sentence to one action and don’t rush the route.

Heisenberg for a longer explanation

Heisenberg has a low, even educational delivery. He can give a longer explanation room, especially when the incident lead has approved several related facts. He measured about 175 words per minute in my clips, close to Derek. The pauses after each line can still be a problem for a short status slot, so trim the script before trying to make a calm read sound urgent.

For the first audition, add Derek to My Voices. Compare the same approved sample on each voice, including the service name and update time. The implementation guide covers the voice ID workflow.

What goes on the factual approval sheet?

Give the incident lead a short review sheet beside the script. It should show the exact sentences proposed for narration, the source for each factual claim, the audience, the publication channel and the time that the update is meant to describe. Mark each sentence as confirmed, pending or removed. Ask the reviewer to flag assurances the evidence cannot support.

The UK NCSC’s 2024 guidance discusses adaptable templates, named spokespeople and backup communications. It also cautions against speculation and premature reassurance. The New Zealand NCSC guidance recommends explaining impact for the relevant audience, being open about unknowns, planning update frequency and using familiar backup channels. Use the points that fit your communications process; neither page supplies facts about a particular business incident.

Record the approver and approval time for the exact wording. A new time, service impact or customer instruction creates a new review. The voice choice doesn’t grant approval, and a clean take cannot validate the script.

How do I keep each revision traceable?

Save the approved script and audio under the same version label. A simple record can include a version number, approval timestamp, publication timestamp, script filename, audio filename and caption filename. When a sentence changes, create a new version and send that text back through the same approval route. Keep the prior published version available to the communications team so they can identify what customers heard.

Make the narration and on-screen text agree. Read the time, date and service name as they appear in the caption file. If the business edits the status page after recording, check whether the audio now points to an old time or an outdated route. Do not quietly replace a file while leaving the old caption or transcript in place.

What should a fictional rehearsal update sound like?

This original sample describes a fictional private business and a fictional customer portal. It makes no claim about Ogi Networks, Data Cymru or any real incident.

This is a service update from our company. Our customer portal is unavailable while the team reviews an interruption. This concerns the portal only. The team is still checking what happened, so a restoration time is unconfirmed. Customers can use the usual support route for order questions. We will post another update at four thirty p.m. Central Time on Tuesday. Check the status page for the current version.

At 68 words, this is about twenty-three seconds with Derek, who measured about 180 words per minute in my clips, before pauses. The time and service status are fictional rehearsal details. Replace them only with facts approved for the actual update.

What should I check in the final audio and captions?

Listen to the exported file from beginning to end at the volume and on the device customers are likely to use. Check that the service name and update time are clear, the spoken words match the approved text and captions, and the last sentence points to the intended channel. Have a second person follow along with the caption file. Fix a mismatch in the source and export a new version; don’t patch only one format.

These checks assess the recording and its text. The incident lead remains responsible for the facts and the decision to publish. For a brief read, I would start with Derek. If customers need one clear next action, try Dexter. I own the voices listed here and may receive ElevenLabs compensation when they are used.

Related guides

Other jobs where I use the same voices.

Nothing playing

Press play on any voice

↑ ↓ moveEnter openShift Enter playEsc