We extensively evaluated a data hiding algorithm for stereo audio signals which embeds data using the polarity of the echoes added to the high-frequency channels, which we have previously proposed. Its performance was also compared to conventional data hiding using spread spectrum, and those using echoes with different delays. Raw embedded data was detected with little or no errors for added noise at 20 dB SNR and above, or with MP3 coders, although the spread spectrum method showed almost no errors at all. However, sample rate conversion and random bit cropping were shown not to affect the embedded data with the proposed method, while other methods, including spread spectrum, showed significant amount of errors. The embedded audio quality test using the MUSHRA standard method resulted in little noticeable degradation, far better quality compared to other hiding methods.