Day 5: Intellectual property and voice likeness

Welcome to Day 5! Today we’ll explore intellectual property and voice likeness in relation to artificial intelligence. 

Today’s activity was written by Roxanne Peters (Senior Creative IP Education Specialist) and Roshni Bhagotra (Senior Digital Learning Coordinator).

Intellectual Property (IP) is a set of international legal rights that protect the financial and reputational value and interests of creative and innovative practice within society. This ranges from pharmaceutical and agricultural industries to the creative outputs of musicians, filmmakers, artists and designers and other forms of popular culture. Types of IP you may have heard of are copyright (creative works) and trademarks (brand and reputation). Other related rights include passing off, personality rights and performer’s rights, all connected to protecting self-identity and reputation.  

IP identifies the dynamics between authorship, ownership and the way people’s investment is recognised (or the injustices or where it has not been), centering IP in conversations about responsibility and accountability is key to ensure AI is used for societal good. As well as to better understand the potential benefits of future AI possibilities.  

An area which is becoming more nuanced and complex legally and ethically is the use of AI Tools such as Eleven Labs and Microsoft’s Azure AI Speech to mimic voice likenesses, often termed as “voice cloning” or “synthetic voice generation.” 

All citizens should have the right to decide how they are represented, and recent cases from voice actors and public figures highlight the argument for legal clarity. For more information, please watch singer and performer FKA Twigs from 20:25 until 25:39 where she talks about intellectual property, misappropriation and advocates for the US “NO FAKES’ act, proposed to prevent non- consensual AI use of artists’ voices, images and likeness.  

In contrast, musicians such as Grimes are inviting others to clone her voice using artificialiIntelligence in order to create new songs. 

Artists have used AI to address issues around ownership, authenticity and ways to democratise the access and use of voice likeness. Holly Herndon (who we also mentioned on Day 2) created an AI voice instrument and website called Holly+ where anyone can upload audio and receive a download of the audio sung back in the artists’ voice. Holly+ was created in collaboration with Never Before Heard Sounds which made machine learning instruments and expressive audio tools. It uses a decentralised model, designed to democratise ways to manage content and allow others to create art using Herndon’s voice. 

Herndon currently has an exhibition with Mat Dryhurst at the Serpentine Gallery in London which explores their vision for collaborative artmaking in the age of AI. (The exhibition closes on 2 February 2025.) 

Like video and other digital formats, sound can be useful and, in some cases, essential for staff and student learning scenarios, as well as for creating accessible and inclusive educational materials. Using elements of your own identity with technology to capture and disseminate education is a common practice, which may be met with different terms and preferences from different people, staff and students, depending on their relationship with their personal assets.   

Activity 

Using the Holly+ voice tool, upload an audio file of you speaking. For inspiration, you could record a reflection of your morning. Describe your journey to work (even if it’s from the kitchen to your home office desk) and anything you experienced along the way. Or, you can upload a short description of an idea for an educational activity you might run. 

  1. Record a short file of you speaking. You can use the ‘Sound Recorder’ application in Windows or on Mac or use the voice note/memo application on your phone. Some apps will save your sound as an m4a file which you’ll need to convert to an mp3. If you don’t have another preferred conversion tool, try using Cloud Convert to do so.
  1. Go to the Holly + website.
  1. Upload your file into the upload area by selecting the Click or Drag to Upload File button or drag and drop your file into the website. 
  1.  The Holly + website will start to generate your audio. Once its completed you can select play to preview the sound or download it direct from the website. 

Important notes from Holly+  

  • There is an upload limit of 5 minutes for audio files  
  • Holly + doesn’t retain a copy of any original audio uploaded to the tool, but the website may archive copies of the output audio. 
  • You are free to use the material generated as you see fit, and accreditation is always appreciated. 

Alternative tools 

You might want to try one of these tools instead:  

Discussion

Join us in the Teams space to share your responses and the sounds you’ve generated. If you don’t have access to the space, email us at teachingexchange@arts.ac.uk for the attention of Hannah. 

  • What are the ethical implications of replicating someone’s voice without their consent?  
  • Does choosing an AI version raise ethical or cultural concerns? Does this differ in an educational context? 
  • Do you think it is ethical to use the voices of deceased artists? If so, can you explain why? 
  • Would you be comfortable giving consent for someone to reuse your voice? If so, why and in which circumstances? 
  • Should voices be protected in the same way your image is protected?   
  • Should our voices be considered as a type of ‘property’?  
  • How would you approach a conversation with colleagues or students about these potential concerns? 
  • Would you feel comfortable if someone else recorded and shared a recording of your voice for educational purposes?