Method and apparatus for processing audio for scene classification

An audio processing apparatus for scene classification, the audio processing apparatus comprising:

• a memory storing at least one instruction;
• and
• at least one processor configured to execute the at least one instruction to:
• obtain a first audio signal corresponding to a first frame;
• extract a first feature vector by inputting the first audio signal to a first neural network;
• obtain a temporal correlation vector representing a similarity between the first feature vector and at least one second feature vector extracted from at least one second audio signal corresponding to at least one second frame that is temporally before the first frame, wherein the at least one second feature vector comprises N second feature vectors, and wherein the temporal correlation vector comprises a first inner product of the first feature vector and the at least one second feature vector, and a second inner product of the first feature vector and the first feature vector;
• and
• classify a scene of the first audio signal by inputting the first feature vector, the at least one second feature vector, and the temporal correlation vector to a second neural network, wherein the second neural network receives an input of a value obtained by concatenating the first feature vector, M second feature vectors from among the N second feature vectors, and the temporal correlation vector, and wherein each of N and M is a natural number, and N is greater than M;
• and
• select, based on a result of classifying the scene, a mode, from among modes of active noise cancellation, and implement the active noise cancellation in the mode selected.

An audio processing method for scene classification, the audio processing method comprising:

• obtaining, from a bitstream, down-mixing-related information comprising a down-mixed audio signal corresponding to an original audio signal and a scene classification result;
• de-mixing the down-mixed audio signal, based on the down-mixing-related information;
• reconstructing an audio signal comprising at least one channel, based on the de-mixed audio signal;
• and
• selecting, based on a result of classifying a scene, a mode, from among modes of active noise cancellation, and implementing the active noise cancellation based on the mode selected,
• wherein the scene classification result is obtained based on concatenation of:
• a first feature vector,
• at least one second feature vector comprising M second feature vectors from among N second feature vectors, and
• a temporal correlation vector between the first feature vector, corresponding to a first frame of the original audio signal, and the at least one second feature vector of the M second feature vectors, the at least one second feature vector corresponding to at least one second frame that is temporally before the first frame of the original audio signal, and wherein the temporal correlation vector comprises a first inner product of the first feature vector and the at least one second feature vector, and a second inner product of the first feature vector and the first feature vector, and
• wherein each of N and M is a natural number, and N is greater than M.