In this paper we propose a novel arrangement for quantizers based on Quantization Index Modulation (QIM) method for data hiding. Due to advantages of non uniform quantization methods, we used the power-law function which is widely used in image processing. The magnitude of host signal is first transformed into the power-law domain. Then the transformed signal is quantized uniformly and the result is transformed back to the original linear domain. We analyzed the performance of the method and the optimum parameter is found. The method is simulated on artificial signals to verify the analytical derivations. The method is also applied to speech signals and its advantage is compared with conventional QIM. The results show that our method has a superior and is robust against common attacks.