Copyright Issues In Synthetic Arabic Voice Narrators For Tourism Guides.
1. Legal Framework in Poland
The main statute is the Act of 4 February 1994 on Copyright and Related Rights. Relevant aspects include:
Copyright in literary works: The script/text of the tourism guide is protected if it meets originality standards.
Protection of performances: Voice narrations, whether human or AI-assisted, may involve neighboring rights if derived from recorded performances.
Computer-generated content: Synthetic voice software is treated under copyright for the software itself.
Moral rights: Attribution and integrity apply to the original script and possibly to distinctive vocal expressions.
EU directives also influence Polish law:
InfoSoc Directive 2001/29/EC – general copyright and communication to the public.
Software Directive 2009/24/EC – protection for voice synthesis software.
DSM Directive 2019/790 – addresses rights and remuneration for digital platforms.
2. Originality in the Script
The synthetic voice itself does not create copyright automatically; the underlying text or narration is the work.
🔹 Case: Infopaq International A/S v Danske Dagblades Forening
Principle: Only the author’s own intellectual creation is protected.
Application: A tourism guide must have creative structuring of content, descriptions, or historical narrative, not just factual listings of sites.
🔹 Case: Painer v Standard VerlagsGmbH
Principle: Even realistic representation of a subject can be protected if creative choices are evident.
Application: Descriptive style, ordering of tour content, and cultural storytelling in the Arabic script can be protected.
3. Protection of Synthetic Voice
Synthetic voices are generated using AI or software. Polish law distinguishes between:
Software code – protected under software copyright.
Recorded voice performances – neighboring rights apply if based on human recordings.
Purely AI-generated synthetic voices – may lack traditional copyright if no human creative input exists.
🔹 Case: SAS Institute v World Programming Ltd
Principle: Software expression is protected; functionality/ideas are not.
Application: Using a synthetic voice engine without a license can infringe copyright in the software code.
🔹 Case: Infopaq International / Painer
Principle: For human-like expression (speech), originality arises from creative input.
Application: AI voices may not automatically gain copyright; human-generated script and prosody decisions matter.
4. Use of Human Voice as Training Data
Many synthetic voices are trained on recordings of professional speakers.
🔹 Case: Deckmyn v Vandersteen
Principle: Reuse of copyrighted material for derivative works requires permission.
Application: Using recorded voices of actors or narrators to train a synthetic Arabic voice without consent can infringe neighboring rights.
🔹 Case: Pelham GmbH v Hütter
Principle: Sampling or using phonograms without authorization can be infringement.
Application: Sampling human voices for AI training in a tourism guide is legally sensitive.
5. Moral Rights in Narration
Polish law protects moral rights, which are inalienable. For narration:
Attribution: AI-generated voices may require credit if modeled on a specific performer.
Integrity: Editing or using a voice in a misleading or derogatory way may violate moral rights.
🔹 Case: Eva-Maria Painer
Principle: Even small modifications that affect reputation or meaning can infringe moral rights.
Application: Altering a synthetic Arabic voice in a tourism app to misrepresent content could trigger claims.
6. Platform Liability
Tourism guides are often distributed via apps or websites.
🔹 Case: GS Media BV v Sanoma Media Netherlands
Linking or embedding infringing content can create liability if the platform knows about it.
🔹 Case: YouTube and Cyando
Platforms are liable if they fail to remove infringing content after knowledge.
AI-generated narration distributed via platforms without proper license could lead to takedown or legal action.
7. Translation and Derivative Works
Translating a tourism guide into Arabic may involve derivative rights.
🔹 Case: Società Consortile Fonografici (SCF) v Del Corso
Adapting content requires permission from the copyright holder if the new work reproduces the original expression.
Direct translation into Arabic counts as a derivative work.
8. AI Voice and Public Domain Texts
Using public domain texts (historical descriptions, architectural facts) is safe for AI narration.
Original scripts remain protected.
🔹 Case: Levola Hengelo BV v Smilde Foods BV
Works must show identifiable human creativity.
Using factual content alone does not create copyright.
9. Practical Risks for Tourism Apps
Using copyrighted scripts for AI voice generation.
Sampling human voices without permission for training.
Embedding licensed AI voices without proper contracts.
Translating copyrighted content into synthetic narration.
Distributing via platforms without knowledge of licensing obligations.
Modifying voice content in a way that infringes moral rights.
Conclusion
Synthetic Arabic voice narrators for tourism guides involve layered copyright:
Script – must be original to qualify.
Voice engine/software – protected under software copyright.
AI-generated voice – limited protection; depends on human input.
Human samples – require licensing.
Moral rights – attribution and integrity remain enforceable.
Platform distribution – liability applies under EU and Polish law.
Creators must carefully manage licensing for both the script and the voice generation software, and avoid unauthorized use of human voices in AI training.
RELATED Blog
comments