# Responsible AI Voice Workflow

This workflow assumes you are using an AI-generated version of your own voice, with your own consent.

## Recommended approach

1. Record a clean consent and training sample.
2. Use a reputable voice provider that supports explicit consent and voice ownership controls.
3. Generate narration only from scripts you have reviewed.
4. Listen to the full output for accuracy, tone, and unintended emphasis.
5. Keep the final audio project files private.
6. Add disclosure wherever the video is published.

## Consent text to record and store privately

"I am creating an AI-generated version of my own voice for use in my personal educational content. I consent to using my voice for narration of material that I write and approve. I do not consent to this voice being used to impersonate me in private conversations, commercial commitments, endorsements, or messages I have not reviewed."

## Disclosure text

Short version:

"Narration uses an AI-generated version of my own voice, created with my consent."

Long version:

"This video is narrated using an AI-generated version of my own voice, created with my consent. The script was written and reviewed by me. The content is for educational purposes and does not include confidential customer information."

## Provider selection criteria

Choose a provider that offers:

- Explicit consent controls.
- Clear voice ownership terms.
- Ability to delete voice models.
- Commercial/public publishing rights.
- Downloadable audio output.
- Controls to prevent unauthorised voice use.

## Production tips

- Generate short sections, not the whole video in one pass.
- Keep sentence length moderate.
- Add pronunciation hints for Azure, GenAIOps, APIM, Entra, and telemetry.
- Use a slower speaking pace than a live conversation.
- Keep a human review step before publishing.