In this research, we analyze the audio watermarking system by combining adaptive wavelet with SVD in various BCH Codes. The host audio signal is embedded by a watermark signal in the form of an image signal accompanied by several attacks on the watermarking system. The results show that ODG and SNR values for DWT level 1 are higher than level 3, that can reach ODG for BCH code 15,11,1 level 1 = -0.58 and BCH code 15,7,2 level 1 = -0.68 on LSC attack type; SNR for BCH code 15,11,1 level 1 = 32.6 dB and BCH code 15,7,2 level 1 = 31 dB on LPF attack type. While for watermark image BER parameter, the result of the research shows that BER level 1 is worse than level 3, and BCH code 15,7,2 has better BER values than those of BCH code 15,11,1. In our system, we have high security watermark image at BCH 15,7,2 level 3 where the BER value is the lowest among other parameters.