This paper proposes a novel speech watermarking algorithm using wavelet transform and BCH coding. The characteristics of the proposed algorithm are as follows: 1) a more steady embedded code strategy adopts to resist the compressed attack more effectively; 2) the uses of multi-resolution property of wavelet transform and the BCH error correcting code to improve the robust of watermark; 3) the watermark will be embedded into the low-, middle-, or high-frequency components by adaptive algorithm according to human perceptual critical bands; 4) the proposed scheme can extract the watermark without using the original speech signal. Experimental results show that the proposed watermarking algorithm is inaudible and robust against G.711, G.726, and GSM 6.10 speech coding standards.