Abstract: The correlation between arc acoustic signals and droplet transfer in aluminum pulsed MIG welding was studied.The arc sound signals under different droplet transfer were acquired by the designed experimental system and analyzed by using the metheds of the wavelet de-noising, the power spectral density and the power spectral estimation of auto regressive and moving average.The analysis results show that the arc acoustic signals have different characters under short-circuit transfer, global transfer, projected droplet transfer and spray transfer, so the pattern of droplet transfer can be recognized by arc acoustic signals.These results have an industrial impact on controlling the stability of droplet transfer of aluminum pulsed MIG welding in real-time.