An audio communication based electronic signature system and a method thereof are provided. The system includes: an audio smart key device including an interface module, a data processing module, a smart key module and a power module; and a terminal including an audio output interface, an audio input interface and an application unit. The interface module of the audio smart key device communicates with the terminal in the form of audio signal and it can perform operations of A/D conversion and modulation and demodulation on the audio signal. In the signing method, the terminal sends an audio signal including important data to the smart key device, the audio smart key device receives and decodes the audio signal, demodulates the decoded data and decrypts the demodulated data, signs the decrypted data, encodes, modulates and encrypts the signature result subsequently, sends the crypted result to the terminal, where the encrypted result is sent to the server for signature verification.