Technology

Performance Capture (PCAP)

The terms MOCAP and PCAP are often used interchangeably and they apply to voice, body, face, and motion depending on the job. PCAP work was initiated from two performance areas, Stunts and Voice Acting. It is not a performance category by itself; it is more of a technical aspect of several performance subcategories.

PCAP has no standard definition in the the industry yet, but here is an example of what it could be:

"Performance where motion, face, body, voice, or puppetry are captured and translated into a digital model that can be manipulated beyond the session performance and displayed within another medium such as film, television, animation, or interactive media."

PCAP technology can include, but is not limited to:

Optical systems (Active & passive)
Inertial systems
Magnetic Systems
Mechanical Systems
Head Mounted Camera (HMC) with voice recording

PCAP performers can be hired as:

On Camera Theatrical/TV Principal, stunt, or Background performers
Animation Principal performers or atmospheric voices
Interactive Principal, stunt, or Background performers

Artificial Intelligence & Synthetic Voice

As the use of Artificial Intelligence (AI) technology continues to evolve and grow across the entertainment industry (and technology at large), opportunities to exploit and harm performers also multiply. Most performers exploring AI performance replication work are unaware of or do not fully understand their rights regarding employment contracts. As synthetic voice work now covers much uncharted territory, it’s imperative that we collaborate to create and amend laws and contracts to protect both laymen and professional performers against deep fakes, improper use and exploitation of recorded performances.

Goals:

To create a standardized definition of this type of work, implement protections for performers, and do educational outreach to the creators and users of AI technology to foster fair and equitable work for everyone.
To actualize legal preservation of vocal likeness and performance integrity, and identify the industry-wide risks of AI voice technology around vocal identity misuse and/or malpractice.
To educate performers and their representatives around the future implications of single or multi-use vocal replication software and/or AI automation, and amend problematic contract terms and language before they become industry-wide.
To create a standardized contract or set of contract clauses that properly protect against abuse and establish a baseline for acceptable rates.

Contracts for performers should include:

Safe Storage and tracking of the performers voice, likeness, performance and all products created from them.
Explicit limits on use of the results and proceeds of a synthetic model.
Ability for the performer to consent to use.
Ability to opt out or term limits for synthetic model use.
Appropriate payment for use.
Clearly denoted exclusivity.

To support the effort to educate the industry on the needs of performers, visit:

www.vocalvariants.com

A Layperson's Guide on How to Create a Synthetic Voice

Training data is required to teach the AI. This data comes in the form of recorded audio files. The accuracy of the voice clone depends on how much audio is recorded and how clean the files are.

The AI machine learns to string phonemes, or sounds, together to create words and sentences.

A neural network takes the ordered set of phonemes and transforms them into a set of spectrograms, or visual frequency pictures.

A neural vocoder then translates these spectrograms into speech waveforms, delivering a natural sounding replication of a voice.

Text to speech (TTS) uses text to generate synthetic speech, often with predeveloped voices.

Speech to speech (STS), uses audio from a person’s voice to create a custom voice. It has the capability to transform a voice into different genders, dialects, and languages if it has the necessary data.

Voice Services Contract Details

Here are things you should look out as you negotiate a contract for an AI voice:

***** Always have a legal representative review contracts*****

Understand what service is covered; is it Text-to-speech (TTS) or Speech-to-speech (STS)?

There should be a payment structure for recording the audio files that feed the AI machine. In union-speak, this is called "session fees".

Use of the audio assets developed by the AI, usually called "results and proceeds" should be explicitly designated.
- Understand if you are providing an AI voice for specific use cases, excluding all others; or providing an AI voice for all use cases, excluding specific ones.
- Determine where the results and proceeds may be geographically used or limited.
- Determine if the you are excluding the use of your audio files for development or training of AI machines.
- Determine if you will allow your AI voice to be used tp develop composite voices.
- Determine what the usage payments will be for each use case of the results and proceeds.

There should be a term of the contract; it should not last forever (in perpetuity)

In the US, it is standard to give intellectual property (IP) rights to the producer. This may include terms like "moral rights", "new exploitation rights", or "transformative rights". Understand the definitions so you are protected against use without the ability to consent of be compensated.

At the end of the term of the contract, all rights should return to the performer.

Understand how the results and proceeds of the AI voice will be tracked, so you can know, consent, and be paid for all usage.

Understand if you will be held exclusive to the company and for how long.

Understand what happens when the results and proceeds are sold for use by the AI company. This is usually covered in a legal "transfer and assumption of rights statement".

Definitions

Active Consent

Consent where the performer (or their representatives) must sign waivers and actively agree to a condition. All other use is excluded.

Passive consent

Consent where a performer agrees to all use as a default unless they actively exclude conditions.

Technical Definitions

Artificial Intelligence

Computer systems able to perform tasks that normally require human intelligence, such as visual perception, speech recognition, decision-making, and translation between languages.

Text-To-Speech (TTS)

Technology that enables text to be converted into speech sounds imitative of a human voice.

Speech-To-Speech (STS)

Technology that enables human speech to be converted into a different vocal print imitative of the human voice.

Types of AI Voice

Voice Double or Voice Clone

A synthetic, adaptable copy of a person or licensed character's voice.

Composite Voice

A synthetic voice that is made up of multiple people's vocal tracks.

Fully Synthetic Voice

A synthetic voice that is not based on a human being.

The exclusive legal right, given to an originator or an assignee to print, publish, perform, film, or record literary, artistic, or musical material, and to authorize others to do the same.

Intellectual Property (IP)

Creations of the mind, such as inventions; literary and artistic works; designs; and symbols, names and images used in commerce.

Moral Rights (droit moral)

There are 2 types of moral rights under copyright law:

1. Right of Paternity

The author’s right to have their name on a work, to use a pseudonym and to remain anonymous.

2. Right of Integrity

The right of the author to object to any changes to their work that may harm their reputation as an author.

New Exploitation Rights

The right to use IP in all media, formats, modes of transmission, and methods of distribution, dissemination, exhibition or performance now known or that may be developed in the future; referred to as “New Exploitation Methods”.

Transformative Work (Rights)

A work that makes use of copyrighted material, but does so in a way that the resulting work is deemed original.