The Use of Audio Fingerprints for Authentication of Speakers on Speech Operated Interfaces

In a multi-speaker and multi-device environment, we need acoustic fingerprint information for authentication between devices. Thus, in these kinds of environments, it is crucial to continuously check the authenticity of speakers and devices within a short duration since different speakers could join or leave the environment. In this work, we propose the provision of different levels of authentication to different speakers in a multi-speaker multi-device environment using acoustic audio fingerprint information. Firstly, the audio fingerprints are extracted continuously every few seconds. Then, the extracted fingerprints are passed to a speaker recognition module which checks if the fingerprint is enrolled for that particular environment or not. Finally, the proper level of authentication is provided for each speaker. Our experimental results on Voxceleb-1 dataset show that acoustic fingerprints can be successfully used for authentication purposes in a multi-speaker multi-device environment.