In this paper, we propose a low-complexity model-based single-channel audio separation approach. The proposed method presents three certain advantages over previous methods: I) replacing commonly used linear masks like Wiener filtering by a proposed non-linear one, we show that it is possible to lower the crosstalk of the interfering source often occurring in a mask-based method while recovering the underlying signals from the observed mixture. Using nonlinear masks establishes a tradeoff between acceptable level of interference and low speech distortion, 2) as a post-processing stage, we use phase synchronization technique to enhance the perceptual quality of the re­synthesized signals, and 3) the proposed method is based on vector quantization {VQ) codebooks. Hence, the complexity is lower than previous GMM-based methods. Through extensive experiments, it is demonstrated that the proposed method can achieve a lower signal-to-distortion ratio (SDR). According to our listening experiments and according to the Mean Opinion Score (MOS) results, it is confirmed that the proposed method is able to recover separated outputs with a higher perceived signal quality.