DEFINITIONS for VOICE ACTORS
Voice actors today face a credibility challenge: how do you prove your work is authentically yours when AI can clone voices instantly? This standard draws a clear line between human performance and machine generation.
QUICK REFERENCE
-
VerifiedHuman label requires VH3 or higher on the Human-AI Spectrum
-
Essential question: "Who gave voice to this sound?"
-
Acceptable AI use: Noise reduction, audio cleanup, technical post-processing
-
Not acceptable: AI-generated or AI-cloned voices you claim as your performance
-
Human voice remains distinct from AI simulation
FOREWORD
The most important definition in the standard is vocal work: the selection and arrangement of essential vocal elements in a performance to be heard—narration, dialogue, or voice-over that has been performed, recorded, or broadcast.
This human creativity standard helps voice actors prove their work is not AI through human-made content certification that validates essential vocalization.
Voice acting is unique. Unlike writers, voice actors typically don't write their scripts—they perform them. The craft lies in bringing words to life through vocal performance. Voiceover must be heard to be experienced; even closed captions can only reference the words, not the inflection, emotion, and nuance that voice actors provide.
The human voice remains distinct from AI simulation. Modern voice production involves technical processes in which AI tools may be employed—from studio recording enhancement to post-production processing—but these tools support, rather than replace, the voice actor's performance. Our spectrum helps voice actors clearly communicate how their work relates to these AI-assisted production elements while maintaining the fundamental distinction between human- and machine-generated voices.
HUMAN-AI COLLABORATION LEVELS
Our 5-point Human-AI Spectrum provides AI transparency certification and human authenticity verification for voice actors. It shows exactly how much of the work is human-performed versus AI-assisted in production. All VerifiedHuman labeled work must meet or exceed Level 3 (VH3), meaning the voice actor performed the vocal work.
VH5 - >99% Human-Performed (<1% AI-Generated)
-
Vocal content performed and produced entirely without AI assistance
-
Includes unique vocal nuances, emotional depth, and personalized expression
-
No traces of automated processes in production
VH4 - Mostly Human-Performed
-
Predominantly human-performed with minimal AI assistance in production
-
May use AI for technical enhancements (noise reduction, audio cleanup)
-
Human vocal expression dominates
VH3 - Balanced Human-AI Collaboration
-
Equal partnership between human performance and AI-assisted production
-
Harmonious integration of human voice and AI-enhanced technical elements
-
Significant human performance input maintained
Work below VH3 is not eligible for the VerifiedHuman label.
In the spirit of this standard, the question remains: Did you perform it, or did AI generate it? We believe the voice actor's intention to create and present vocal work as their own should always remain with them. Therefore, we urge all voice actors to authenticate their work in the performance process to ensure that their final product is unique and distinctive.
We acknowledge the hard work of skilled professionals - women and men who have honed their skills and mastered this craft in various languages. We appreciate your human voices and are grateful to be able to help you mark them as appropriately yours.
DEFINITIONS
Here are definitions of words and ideas used in the Standard for Voice Actors.
Standard
A specific description of human behavior
Voice Actor/Voiceover Artist
The narrator/performer of vocal work
Represent
Present, share, or show work with others
Vocal work
The narration or production of essential vocal elements for live performance, broadcast, or recording
Team
A group of people working together
Intellectual property
A work or invention that is the result of creativity, such as a manuscript or a design, to which one has rights and for which one may apply for a patent, copyright, trademark, etc.
Essential
The fundamental elements or characteristics of something
Essentially vocalized
When a voice actor narrates or produces essential vocal elements in a meaningful way to create vocal work (see Essential Vocalization below)
Essential vocal elements
Sounds made by the vocal tract–can be heard as talking, singing, laughing, crying, screaming, humming, mumbling, or shouting
Human
Noun: a human person; adjective: from a human person
Generative AI
Shorter definition–machines that create novel (new) content
Longer definition–generative artificial intelligence (AI) describes algorithms (such as ChatGPT) that can be used to create new content, including audio, code, images, text, simulations, and videos.
Machine learning
"Systems that act like humans"
Other generative processes
Other processes involving AI or machine learning to create novel (new) content
OTHER DEFINITIONS
Other definitions of words and ideas related to the Standard for Voice Actors are here.
Accent
The prominence of a syllable; characteristic pronunciations of one spoken language that can be heard in another
Artificial Intelligence (AI)
A field that combines computer science and robust datasets to enable problem-solving. It also encompasses the machine learning and deep learning subfields, which are frequently mentioned in conjunction with artificial intelligence. These disciplines involve AI algorithms that aim to build expert systems capable of making predictions or classifications from input data.*
AI language modeling (or Large Language Modeling (LLM))
Systems that can use natural language text from large amounts of data. Large language models use deep neural networks, such as transformers, to learn from billions or trillions of words and to produce texts on any topic or domain. Large language models can also perform a range of natural language tasks, including classification, summarization, translation, generation, and dialogue. Some examples of large language models are ChatGPT, Claude, and Gemini.**
Audio output
The production or reproduction of sound, live or recorded
Emphasis
Special and significant stress of voice laid on particular words or syllables, stress laid on particular words by means of position or repetition
Interpretation
Way of understanding or explaining the meaning of something
Language
Communication by voice in the distinctively human manner, speech; a body of words and the systems for their use common to people who are of the same community or nation, the same geographical area, or the same cultural tradition
Inflection
The rise and fall of pitch in the voice; changes in tone used to convey meaning, emotion, or emphasis. Inflection helps distinguish questions from statements, conveys sarcasm or sincerity, and brings emotional depth to vocal performance.
Morphological vocal processing
When the features and structures of a vocal sound's characteristics are described and processed as words
Pause
A temporary stop or rest
Personal, original idea
An idea representing a specific human being's unique insight or experience in the world
Pitch
Height or depth of a tone or sound, depending upon the relative rapidity of the vibrations by which it is produced
Pronunciation
Producing the sounds of speech, including articulation, stress, and intonation, often concerning a standard of correctness or acceptability
Rate/Speed
A measure, quantity, or frequency, typically one measured against some other quantity or measure, rate of motion, or progress
Representation and description
Representing a sound in a way that can be analyzed by a computer, like numbers or words that have been assigned to specific sounds and features of recorded sounds
Sound
Vibrations transmitted through the air or other medium experienced through hearing
Text-to-voice generation
Also known as text-to-speech (TTS), using written or spoken words to prompt an AI generator to create vocal work
Timbre
The characteristic quality of sound, independent of pitch or volume, from which the manner of production can be inferred, dependent on the relative components of resonant frequencies
Tone
A sound considered concerning its quality, pitch, strength, and source; the quality or character of a sound
Values
Principles or standards of behavior
Values-based
Relating to principles, values, or ethical assumptions that motivate human behavior
Voiceover
The voice of an offscreen narrator, announcer, speaker, or reader, as in a commercial, using such a voice
COMMON USES
Here are commonly accepted ways AI is used in voice production.
Audio cleanup and noise reduction uses AI to remove background noise, breathing sounds, mouth clicks, and environmental interference from recordings. This is widely accepted.
Pitch and timing correction uses AI to adjust vocal pitch or timing while maintaining the natural quality of the human voice. Standard in professional production.
Audio mastering and enhancement uses AI to balance levels, apply EQ, compression, and other effects to improve overall sound quality. Industry standard.
Script generation and prompts where AI helps generate ideas or draft scripts that the voice actor then performs in their own voice. The performance remains human.
Dubbing and lip-sync timing uses AI to help match vocal timing to visual content across languages. The voice actor still performs all dialogue.
INTERPRETATION
Here are questions voice actors commonly face when working with AI tools.
Q: Can I use AI to clean up my recordings and still qualify for VerifiedHuman?
ACCEPTABLE: Using AI for technical post-processing (noise reduction, EQ, compression, mastering) while you performed the entire vocal work. AI enhanced your recording; you created the performance.
GRAY AREA: Using heavy pitch correction or vocal processing that significantly alters your natural voice. Ask yourself: Is this still recognizably my performance?
NOT ACCEPTABLE: Using AI voice cloning or synthesis to generate vocal performances. If AI generated the voice, you didn't perform it—even if you wrote the script or provided the voice sample.
The question remains: Who performed it?
REAL-WORLD SCENARIO: Audiobook Narration with AI Enhancement
Sarah narrates a 10-hour audiobook. She uses AI to:
-
Remove background noise and mouth clicks (audio cleanup)
-
Normalize volume levels across chapters (mastering)
-
Reduce echo in her home studio (acoustic correction)
-
Apply light compression for consistency (production enhancement)
Sarah performed every word, character voice, and emotional inflection herself.
VERDICT: VH4-VH5 (Mostly/Entirely Human-Performed)
WHY IT QUALIFIES: AI assisted with technical post-production, but Sarah performed all vocal work. She is the essential vocalizer.
ESSENTIAL VOCALIZATION
In vocal performance, artists use their minds and distinctive physical and verbal traits to bring human expressions to life, crafting unique soundscapes. These come alive in movies, radio, audiobooks, and more. We establish essential human vocalization with a simple question.
The essential question of vocalization is: Who (or what) gave voice to this sound?
RATIONALE
We hear and experience spoken vocalizations, either human voices or voices simulated by a machine.
The result is vocal work when a human's voice can be experienced live or in recorded media.
ASSUMPTION OF ESSENTIAL HUMAN VOCALIZATION
If a human (and not a machine) gives voice to words and sounds, then a human is the essential vocalizer.
Voice Actors use VerifiedHuman's human-made content certification to prove their work is not AI-generated. Our human creativity standard, established in April 2023, provides AI transparency certification and human authenticity verification for all vocal work. Free to join. 190+ creators certified worldwide.
