Description:
Identifies AI generated audio through DNN based models and disects audio to examine clues of AI manipulation that are difficult for humans to detect. Puts audio data through “spectrogram-like time-frequency embeddings”, which goes through a classification process. Probability ranging from 0 to 1, with 1 insuating that there is a high chance of the audio being fake. [3]
Advantage:
Over 98% accuracy [3] Isolates voices for better accuracy [4]
Limitations:
The free option is limited to 1,000 seconds each month. [4] Not publicly available for development
References:
[3] Ahmed, Z. (2024, February 6). Top 3 Deepfake Detection Tools of 2023. Resemble AI. https://www.resemble.ai/top-deepfake-detection-tools/
[4] Resemble AI – Detect – DetectorTools.ai. (n.d.). DetectorTools.ai. https://detectortools.ai/tool/resemble-ai-detect/
How to Access:
Purchase Service
(Request a Demo)
Resources:
Last Accessed:
11/16/2024