An audio data interpolation apparatus and method for creating interpolated data corresponding to an error position in audio data using a filter having a filter characteristic that corresponds to a feature amount of the audio data, in accordance with at least data pieces before the error position of the audio data, and replacing the data portion at the error position of the audio data with the interpolated data.