Estimation device, learning device, estimation method, learning method, and recording medium

A learning device comprising:

• a memory;
• and
• processing circuitry coupled to the memory and configured to:
• receive an input of a training input audio signal that is an audio signal in which sounds from a plurality of sound sources are mixed, and an input of supplemental information, and output an estimation result of mask information that identifies a mask for extracting a sound of any one of the sound sources included in an entire or a part of a signal identified by the supplemental information, the signal being included in the training input audio signal,
• cause a neural network to iterate a process of outputting the estimation result of the mask information,
• update parameters of the neural network, based on a result of a comparison between information corresponding to the estimation result of the mask information obtained by the neural network, and information corresponding to correct answer mask information given in advance for the training input audio signal,
• cause the neural network to output an estimation result of the mask information for a different sound source, by inputting a different piece of the supplemental information to the neural network at each iteration, wherein
• the supplemental information is information for identifying a remaining signal resultant of removing a signal of a sound source corresponding to the estimation result of the mask information obtained in a past, the signal being included in the input audio signal, from the input audio signal.

A learning device comprising:

• a memory;
• and
• processing circuitry coupled to the memory and configured to:
• receive an input of a training input audio signal that is an audio signal in which sounds from a plurality of sound sources are mixed, and an input of supplemental information, and output an estimation result of mask information that identifies a mask for extracting a sound of any one of the sound sources included in an entire or a part of a signal identified by the supplemental information, the signal being included in the training input audio signal,
• cause a neural network to iterate a process of outputting the estimation result of the mask information,
• update parameters of the neural network, based on a result of a comparison between information corresponding to the estimation result of the mask information obtained by the neural network, and information corresponding to correct answer mask information given in advance for the training input audio signal,
• cause the neural network to output an estimation result of the mask information for a different sound source, by inputting a different piece of the supplemental information to the neural network at each iteration, wherein the supplemental information is information for identifying a remaining signal resultant of removing a signal of a sound source corresponding to the estimation result of the mask information obtained in a past, the signal being included in the input audio signal, from the input audio signal, and
• update parameters of the neural network in such a manner that the estimation result of the mask information is brought closer to the correct answer mask information.

An estimation method executed by an estimation device, the estimation method comprising:

• receiving an input of an input audio signal that is an audio signal in which sounds from a plurality of sound sources are mixed, and an input of supplemental information, and causing a neural network to obtain an estimation result of mask information that identifies a mask for extracting a sound of any one of the sound sources included entire or a part of a signal included in the input audio signal, the signal being identified by the supplemental information;
• and
• executing a process of obtaining the estimation result of the mask information, iteratively, wherein
• an estimation result of mask information corresponding to a different sound source is obtained in each iteration, by inputting a different piece of the supplemental information to the neural network, and
• the supplemental information is information for identifying a remaining signal resultant of removing a signal of a sound source corresponding to the estimation result of the mask information obtained in a past, the signal being included in the input audio signal, from the input audio signal.