The application discloses a voice data processing method, system and medium, belonging to the technical field of bluetooth audio data processing. The method includes: analyzing the current audio frame to be encoded, and determining the encoding bit rate corresponding to the current audio frame, the encoding bit rate includes a full bit rate, a medium bit rate and a low bit rate, wherein the full bit rate is greater than the medium bit rate, and the medium bit rate The rate is greater than the low bit rate; the number of retransmissions is set according to the encoding bit rate, where the encoding bit rate is positively correlated with the number of retransmissions; the current audio frame is encoded according to the encoding bit rate, and the encoding result is transmitted. According to the characteristics of the audio frame, the present application selects the corresponding encoding rate to encode the audio frame, and sets different retransmission times correspondingly. It reduces the number of retransmissions or even does not retransmit some audio frames for which there is no valid voice, avoiding wasting audio transmission bandwidth, reducing power consumption and delay.