Skip to main content

Why did I get a wrong result on Reality Defender, and how should I interpret that?

Diana Hsieh avatar
Written by Diana Hsieh
Updated over a year ago

Our deepfake detectors achieve accuracies that are among the best in the market. However, we must acknowledge that it is nearly impossible to achieve 100% accuracy on all types of input audio. There are a two main reasons:

  • The problem of deepfake detection is an evolving one, with new and more-sophisticated deepfakes emerging everyday

  • There could be a strong mismatch between our training data and the input audio. For example, we have not (yet) considered background music in our training data, and so our accuracies will not be very high for this type of inputs.

Did this answer your question?