Method and system for correcting infant crying identification

A system for correcting infant crying identification, which is for identifying a crying of an infant, the system for correcting infant crying identification comprising:

• an audio unit detecting a sound around the infant to generate a plurality of audio samples;
• and
• a processing module electrically connected to the audio unit, wherein the processing module comprises:
• a memory, comprising:
• a real result corresponding to the crying of the infant;
• a common model created from the audio samples;
• and
• an incremental model created from a plurality of infant crying features;
• and
• a processing unit connected to the audio unit and the memory and receiving the audio samples, wherein the processing unit converts the audio samples to generate a plurality of audio spectrograms, the processing unit extracts the audio spectrograms through the common model to generate a plurality of infant crying features, the processing unit trains the infant crying features through the incremental model to generate an identification result, and the processing unit judges whether the identification result is correct according to the real result;
• wherein in response to the identification result is different from the real result, the processing unit generates an incorrect result, and the processing unit corrects the incremental model according to the incorrect result;
• wherein,
• a cumulative number of the incorrect results is set to 0;
• in response to the identification result is different from the real result, the cumulative number is accumulated;
• and
• in response to the cumulative number is between 5 and 10, the correcting step is executed.