Audio Digital Signature Algorithm with Tamper Detection

An audio digital signature algorithm based on Principal Component Analysis (PCA) was presented. Wavelet packet decomposition was implemented to audio files, all the terminal coefficients of the wavelet packet decomposing tree were used as the data matrix. The matrix was chaotically encrypted, and then it was Principal Component Analysis that to get the eigenvector. Quantized eigenvector matrix was used as the signature, and the signature was embedded into the primal audio data by using watermarking system, and at the same time ICA technology was introduced to locate the audio tamper. The experiment results confirm that the signature is sensitive to the change of the audio data and can locate the tamper.