Canonical page: https://verifiedhuman.info/verifiedhuman-for-voice-actors
Reviewed library edition: 2026-10-29

# VerifiedHuman for voice actors

Your timing, breath, and interpretation can change how someone hears a sentence. The VerifiedHuman mark helps listeners recognize the human performer making those choices. Start with the [Standard for Voice Actors](https://verifiedhuman.info/voice-actors) and its [definitions](https://verifiedhuman.info/definitions-for-voice-actors).

## Your commitment

> I represent my vocal work to be my intellectual property, essentially vocalized by (a) human(s), and not by generative AI.

The central question is: Who gave voice to this?

## The performance comes from a person

Timing, character, emotion, and interpretation come from the performer. The script can be written by someone else. What matters here is who gave the performance.

Noise reduction, EQ, compression, mastering, and other production tools can help a recording. If processing is substantial, consider whether the result still carries your performance and interpretation.

## A cloned voice is a different thing

A system can generate new speech that sounds like you. Supplying the voice sample or writing the script does not mean you performed that new speech.

AI-generated and AI-cloned vocals do not qualify under this standard. Transforming them does not make them human performances. At VH3, VH4, and VH5, the core performance must be human.

## Be clear about the recording

Describe important production assistance and credit the performer. Use the mark on the qualifying vocal work, without implying that it certifies every other part of a production.

If another audio component’s source is unclear, the AIU (AI-Unknown) tag can record that uncertainty. The marked vocal performance must still be human.

## Worked example: recording your own performance

Imagine a voice actor performs a script and uses AI-assisted tools to reduce noise or help with production. The vocal performance is still the actor’s. Assess the production assistance against the [voice-acting definitions](https://verifiedhuman.info/definitions-for-voice-actors); a human gives the performance at every qualifying level.

A process note could say: “Performed by me. AI-assisted noise reduction was used in production.” Use it only if that describes the recording. Traditional processing and generative assistance are distinguished in the definitions, so the tool’s role matters when choosing the level.

## Worked example: using a clone of your own voice

Imagine a performer supplies recordings to a voice-cloning service, then generates a new spoken performance from text. The generated speech does not qualify as a human vocal performance, even when the voice model belongs to that performer or they edit the output. The [human-performance requirement](https://verifiedhuman.info/definitions-for-voice-actors) still applies.

These are hypothetical illustrations of the current standard, not decisions about named members or a new rule for qualification.


## Read next

- [Read the standard and join](https://verifiedhuman.info/voice-actors)
- [Definitions and examples](https://verifiedhuman.info/definitions-for-voice-actors)
- [Use the mark](https://verifiedhuman.info/brand)