Product Demo Voiceover With AI: A Practical Workflow
Product demo voiceover with AI: script length by pace, a step-by-step workflow, caption checks, and how to update the voice when your product UI changes.
To add a product demo voiceover with AI, write the script first at about 150 words per minute, render the voice, then cut the screen recording to the audio. Keep educational demos between one and five minutes, check whether the voice imitates a real person, and plan captions, which WCAG 2.2 requires for prerecorded synchronized media.
How long should a product demo video be?
Length depends on the job the video does. Wistia's length guide, dated July 1, 2025, sorts videos by goal. Under one minute is for awareness. One to five minutes is for education, and how-tos and product demos fall in that band. Five to 30 minutes is for conversion or nurturing, such as webinars and walkthroughs.
| Length | Wistia's stated goal | Typical use | |---|---|---| | Under 1 minute | Raise brand awareness | Social teasers | | 1 to 5 minutes | Educate or explain | Product demos and how-tos | | 5 to 30 minutes | Convert or nurture | Walkthroughs and webinars |
Wistia also notes that most viewers drop off around the halfway point, so put the most useful information first. Its 2026 State of Video recap, dated April 22, 2026, covers play rates and does not include product demo statistics, so the length bands above are guidance rather than measured results.
How many words to script
Most narrators speak at about 150 words per minute. Breadnbeyond uses the same 2.5 words per second for explainer scripts, and Killer Spots puts a professional conversational read at about 2.5 words per second, slower at 2.2 for a warm read and up to about 3 for fast copy. Dense screens need slower delivery.
| Demo length | Words at 150 per minute (derived) | |---|---| | 30 seconds | 75 | | 60 seconds | 150 | | 90 seconds | 225 | | 2 minutes | 300 | | 3 minutes | 450 | | 5 minutes | 750 |
Voices differ in speed. Microsoft's documentation lists words-per-minute values per voice that range from 139 to 293 across languages, so time a draft render rather than trusting the 150 figure.
Step-by-step: voice a product demo with AI
- Outline the demo in three to five steps, one for each screen the viewer will see.
- Write the script at about 150 words per minute, and count the words in each step.
- Audition voices on the Free plan, which includes 5 saved clips.
- Render a draft and time it against the screen recording. Trim any step that runs long.
- Render the final audio, then edit the screen recording to match it in your video tool. OndaLabs does not edit video.
- Draft captions from the script. Speech to text can transcribe audio at 50 credits per minute, which can save time, but review and time the captions in your video tool.
- Publish, and check the disclosure rules for the platform before you upload.
Script rules that make narration easy to sync
- Name the screen before you describe it, so the viewer's eyes and ears agree.
- Put the action first. "Click Export" is easier to follow than "the export button is in the top right."
- Leave a short pause at each screen change. Punctuation and plain-language delivery direction help here.
- Describe what a button does rather than reading long interface labels word for word.
- Round numbers so they read cleanly, and spell out product names as they should be pronounced.
Captions and audio description
WCAG 2.2 treats captions as a standard for prerecorded audio in synchronized media. Success Criterion 1.2.2 (Level A) asks for captions that also identify speakers and meaningful sounds. Success Criterion 1.2.5 (Level AA) asks for audio description of prerecorded video, but says none is needed when the audio track already conveys all the important information. A demo whose narration describes what is on screen mostly meets the spirit of that test. WCAG is a standard, and legal requirements vary by jurisdiction.
Keeping the demo current
Interfaces change, and a voiceover must change with them. Keep the script in a document with one line per step, so you can re-render only the lines that changed and replace those clips. Re-rendering a three-minute script costs about 2,700 credits on Onda Lite, roughly 1.6 percent of Starter's 170,000 monthly credits, or about 4,050 credits on Onda Flash (derived, as of October 17, 2026).
Localising a demo
A three-minute, 450-word demo costs about $0.14 on Onda Lite and about $0.21 on Onda Flash on Starter, as of October 17, 2026 (derived). To localise it, translate the script and have a native speaker check product names and technical terms. OndaLabs has 37 language pages, and its 30 prebuilt voices work in every supported language. Human narration for 301 to 750 words is $450 to $550 at Voice Crafters, as of the same date, which is a useful budget comparison.
Disclosure: when AI voice labels apply
YouTube's altered and synthetic content page asks creators to disclose content that makes a real person appear to say or do something they did not. It lists cloning your own voice for voiceovers as a minor edit that needs no disclosure, and it exempts unrealistic content. A generic AI narrator reading a walkthrough is not one of the examples, but the page is not exhaustive, so read it before you publish.
The EU's deepfake disclosure rule under Article 50 concerns content that "would falsely appear to a person to be authentic or truthful." A narrator voice is not a deepfake unless it imitates a real person. Check that first.
Where OndaLabs fits
OndaLabs offers text to speech with 30 prebuilt voices, and Onda Lite costs 1 credit per character on every plan. Use text to speech to render narration and dialogue for a customer-and-expert style demo. Custom voices from the Creator plan keep one voice across every video, cloned from a recording you have permission to use or designed from a description. Commercial use starts on Starter at $9 a month, so use the Free plan to draft and check your voices and plan options before you publish.
For more on writing the script, read how to write a voiceover script. For the budget side, read voiceover cost, AI versus human.
Frequently asked questions
How do I add a voiceover to a product demo video?
Write the script first, at about 150 words per minute, then render the voice. Edit the screen recording to match the audio, and time a draft so the narration fits each step. Finish by reviewing captions and the disclosure rules for your platform.
How long should a product demo video be?
Wistia's guide suggests one to five minutes for an educational demo, and five to 30 minutes for a walkthrough or webinar aimed at conversion. Shorter demos hold attention better, but the right length depends on what the viewer needs to learn.
How many words should a demo script have?
At about 150 words per minute, a one-minute demo needs about 150 words and a three-minute demo about 450. Dense screens need slower delivery and fewer words. Time a draft render before you lock the script.
Can I use an AI voice for a software demo?
Yes, if your plan includes commercial use and the voice does not imitate a real person without permission. Keep the narration accurate to the current interface, and re-render the lines that change after each release.
How do I keep a demo voiceover in sync when the UI changes?
Keep the script in a document with one line per step, and re-render only the lines that changed. Then replace those clips in your video editor and check the timing. Re-rendering a three-minute script is a small share of a monthly plan's credits.
Should I disclose that a demo uses an AI voice?
Check the platform's rules and the local law. YouTube requires disclosure for realistic content that shows a real person saying or doing something they did not, and EU rules concern deepfakes. A generic AI narrator on a product walkthrough usually sits outside those examples, but read the policy for your platform before you publish.