Description

A "deepfake" voice is a computer-generated voice which has been artificially created using software that clones the voice of a real person. Deepfakes are created using Artificial Intelligence (AI), and can be used by criminals to impersonate people. This experiment aims to explore how people perceive deepfake voices versus real human voices.

You will be asked to listen to an audio file, then select points in the clip if they suggest to you that the file may be a deepfake. You will also be asked to provide comments and decide whether you think each file is real (produced by a human voice) or synthetic/deepfake.

The survey contains 10 questions, all of the same format.

Terms

Delineating the Deepfake: an Analysis of Listener Perceptions of Synthetic Speech

Study Information
Researcher: Amy Cope
Researcher email: ac2796@york.ac.uk
Chair of LLS Ethics Committee email: ben.gibb-reid@york.ac.uk

What is the research about?
The research aims to explore how people perceive deepfake speech.

What does the study involve?
The survey will contain a series of short audio clips. You will be asked listen to the clip and click if you hear anything which sounds odd, non-human, or gives the idea that the speech may be a deepfake. These clicks can then be annotated to describe exactly what it is that you think was odd or why you chose to click on that word or part of the recording.

What will happen to the data I provide?
The data you provide will be entirely anonymised such that none of the information you provide could be linked back to you and any personal details will be destroyed. The anonymous data will be retained for future research.

This study is run in compliance with the University’s General Privacy Notice and will respect all your rights as described therein.

Do I have to take part?
No. Your participation in this study is entirely voluntary. You may withdraw from the study at any time during the survey. If you wish to withdraw from the study after the completion of the survey, please contact the researcher using the email provided above. Note that after a week of the survey being closed, it will not be possible to withdraw from the study because any personally identifying information will be destroyed at that point.

Participant Consent
By clicking below you confirm that you have read and understood the information above, and that you agree to take part in the study.