Definitions for Voice Actors

At a glance

The threshold: Vocal work must meet VH3 (Human-Led, AI-Assisted) or above to carry the VerifiedHuman mark. At every qualifying level, the vocal performance is human.
The question: Who gave voice to this?
Technical assistance: Noise reduction, EQ, compression, mastering, and audio cleanup can support a performance you gave.
The boundary: AI-generated or AI-cloned vocals do not qualify as your performance, even when they use your script or a sample of your voice.
Heavy processing: Consider whether the result still expresses your human performance and interpretation.
Uncertain origins: Add AIU (AI-Unknown) when the origin of audio material is unclear. The qualifying vocal performance must still be human.

Essential Vocalization

A voice actor gives words their timing, inflection, emotion, and character. The script may have been written by someone else. The performance comes from the person who brings it to life.

Who gave voice to this? That is the question behind essential vocalization. When a person performs the words and sounds, that person is the essential vocalizer. When a system generates them, the sound is a synthetic performance.

The distinction applies to live work and recordings, from dialogue and narration to singing, laughter, and other vocal expression. Captions can carry the words. The voice carries the performance.

When the source is uncertain

You may work with audio components whose human or AI origin is unclear. When that happens, add AIU (AI-Unknown) to your VH level, for example, VH4-AIU.

The tag identifies uncertainty about source material. It does not change the central requirement: the vocal work carrying the mark must be a human performance.

Go deeper.

Open the explanations, examples, and terms you need.

Human performance and AI

The human role behind the standard.

Listeners deserve to know whether they are hearing a human performance. VerifiedHuman certifies human vocal work through a shared standard and the performer’s commitment to it.

Recording and post-production tools can improve a performance’s sound. The standard allows technical assistance while keeping the human voice at the center of the work.

Voice actors may experiment with creative effects and stylized productions. Use the mark on work that meets the standard. It honors the skill, interpretation, and distinctive expression you bring to a performance.

Which level describes your work?

From entirely AI-generated to entirely human.

The Human–AI Spectrum has five levels. Vocal work at VH3, VH4, or VH5 qualifies for the VerifiedHuman mark.

At all three levels, a human performs the vocal work. Transforming a synthetic or cloned voice does not make it a human performance.

VH5 · Entirely Human-Performed

You perform all vocal content without generative AI.
Traditional noise reduction, EQ, compression, and mastering are fine.
The work reflects your voice, interpretation, and emotional expression.

VH4 · Human-Performed, AI-Enhanced

You perform all vocal content.
Generative AI may help with scripts, research, session notes, and technical documentation.
Your voice and artistic choices lead the work. AI supports the workflow without generating the vocal performance.

VH3 · Human-Led, AI-Assisted

You lead and give the vocal performance.
AI may make a substantial contribution to production, including heavy processing or timing adjustments.
The core performance remains human and reflects your voice and interpretation. AI-generated or AI-cloned vocals do not qualify.

Work below VH3 does not qualify for the VerifiedHuman mark.

See the complete Human–AI Spectrum

Using AI in voice work

How the standard applies to AI assistance.

You are responsible for judging whether your work meets the standard. Begin with the performance, then consider what happened to the recording.

Cleaning up your recording

You may use AI for noise reduction, EQ, compression, mastering, or other technical cleanup when you performed the vocal work. AI has helped process the recording. You supplied the performance.

Processing that changes the voice

Heavy pitch correction or vocal processing needs closer consideration. Ask whether the result is still recognizably your performance and whether your interpretation remains audible.

At VH3, significant production assistance is possible, but the core vocal performance must remain human.

Voice cloning and synthesis

AI voice cloning or synthesis generates vocal work that you did not perform. Writing the script or supplying the original voice sample does not change who generated the new performance.

A voice that sounds like you is not necessarily a performance you gave. The question remains: Who gave voice to this?

AI assistance in practice

A closer look at a real creative decision.

An audiobook recording

Sarah narrates a ten-hour audiobook. She performs every word, character voice, and emotional inflection.

She uses AI tools to remove background noise and mouth clicks, normalize volume across chapters, reduce room echo, and apply light compression.

Sarah remains the essential vocalizer. The tools improve her recording. They do not generate the narration.

This example falls within VH4–VH5, depending on the tools used. Traditional processing without generative AI is consistent with VH5. Generative AI assistance must be assessed under the level descriptions above.

Terms used in the standard

The language used in the commitment.

Voice actor / voiceover artist: The performer of vocal work.
Vocal work: Vocal elements produced for live performance, broadcast, or recording, including narration, dialogue, and voice-over.
Essentially vocalized: Performed by a person who gives voice to the words and sounds. The central question is: Who gave voice to this?
Essential vocal elements: Sounds made by the vocal tract, including talking, singing, laughing, crying, screaming, humming, mumbling, and shouting.
Voice synthesis / text-to-speech: A system that generates spoken audio from text, producing a synthetic voice rather than recording a human performance.
Voice cloning: A system trained on recordings of a person’s voice that generates new speech resembling that voice.
Inflection: Changes in pitch and tone that convey meaning, emotion, or emphasis.
Timbre: The characteristic quality of a voice, distinct from its pitch or volume.
Standard: A statement of commitment to specific creative practices.
Represent: Claim, present, or share work as your own.
Team: A group of people working together.
Intellectual property: Creative work to which someone has legal rights, such as copyright or trademark.
Essential: The fundamental elements or characteristics of something.
Human: A person, or something originating from a person.
Generative AI: AI systems trained on existing content that generate new text, images, audio, code, or video in response to prompts.
Machine learning: Algorithms that learn from data to make predictions or generate content.
Other generative processes: Other AI or machine-learning processes that create new content.
Artificial intelligence (AI): A field of computer science concerned with systems that perform tasks such as learning, reasoning, and generating content.
AIU (AI-Unknown): A tag added to a VH level when the human or AI origin of source material is uncertain.

Ready to join?

Read the Standard for Voice Actors and make the commitment behind the mark.

Read the standard and join

Free to join. Pay what you can.

Definitions for other fields

Writers Visual Artists Musicians Content Creators

Human–AI Spectrum

Definitions forVoice Actors