AI voiceover and narration

What it does

A DemoVexa project can use AI narration from a connected ElevenLabs, Sarvam or Deepgram provider, your own recording, or no voice with captions and music. ElevenLabs supplies word-level timings; Sarvam and Deepgram use estimated word highlighting.

How to use it

Choose the voice option in the project, select or record the narration, listen through it, and regenerate after script or voice changes. The animation is built around the voiceover; check timings in the preview, especially with estimated word highlighting.

• Voice step and narration options

Where it helps

Narration can help explain feature demos, onboarding videos and tutorials. A human voice or captions-only option is also available when synthetic narration is not the right fit.

• Feature demo
• Customer onboarding
• Tutorial / how-to

A practical example

For a short onboarding clip, compare a connected voice, a user's recording and the captions-with-music option before choosing the version that fits the audience.

What to check before sharing

Listen for names and pronunciation, verify language and voice rights, then regenerate narration and rebuild timing after script or voice edits.

Compare the final video with the current product, confirm the selected plan and provider limits, and review the preview at each format before export.

Plan a small, verifiable first use

Start with one outcome this feature should help you prepare, and decide who will judge whether it is ready. Gather only the current product context the task needs: the relevant local app folder or guide, the intended audience, and any voice, brand or format choice that affects the result. Use fictional data and keep credentials, private customer details and unreleased claims out of the brief.

Build a short representative example before preparing a larger set. Review the source and settings, then compare the script and scenes with the product behavior they describe. If the feature depends on a provider, language, plan or destination, state that dependency plainly so the reviewer can judge the right version of the output.

Review changes and pass the files along

When the first preview is ready, check the result in context: listen to the full narration, inspect captions against the audio, and preview each selected aspect ratio. Ask the product owner to verify behavior and labels; ask the publishing owner to check readability, duration and destination requirements. A successful setting or generated draft does not itself confirm that the video is accurate or ready to publish.

For handoff, include the approved MP4, subtitle files when available, the intended placement and the product version used. Keep a reviewer and review date with the asset. Revisit it when the underlying product, provider options or publishing requirements change, then rebuild and review the affected output before replacing an existing video.

Decide whether this feature fits the job

Use this workflow when a reviewed, animated explanation is useful and your team can check it against current product context. If the viewer needs to see a live-only behavior, a real person demonstrating hardware, or an interactive path they can explore themselves, consider whether a screen recording, live demonstration or interactive guide would answer the question more directly. Choose based on what the audience needs to observe.

For this feature, define a simple acceptance check before production: name the visible output that must be present and one condition that would make the draft unsuitable. That could be a required caption track, a verified provider voice, a readable mobile composition or a release-approved screen. Share that check with reviewers so feedback leads to a clear edit or approval decision.