Most watermarking schemes in literatures are usually implemented by embedding a mark in a host media itself for intended purposes. The existing techniques for watermarking are confronted with the problems of perceptible quality degradation and the inherent conflict between imperceptibility and robustness, which introduced by the watermark embedding. In this paper, we propose a novel audio zero-watermarking scheme for content authentication. First the audio is divided to frame by fixed length and then low-frequent and high-frequent components are obtained by three-level lifting-based wavelet transformation in every frame. Secondly the audio frame is approximately represented as a product of a base matrix and a coefficient matrix using non-negative matrix factorization (NMF). Finally the sum of each column in the coefficient matrix is calculated which is quantized to produce a feature vector, and then the copyright information is obtained by calculating the watermark and feature vector. Experiment results show that the proposed scheme is robust against Mp3 compression and secure.