This technique relates to an earhole attachment-type sound pickup device, a signal processing device, and a sound pickup method which make it possible to achieve high-S/N-ratio voice pickup in which the influence of noise is reduced without using noise reduction processing such as the SS method. As an earhole attachment-type sound pickup device, a microphone that picks up spoken voice is provided in a space that is substantially hermetically sealed from the outside and that is in communication with the external auditory meatus of a wearer, and spoken voice propagated via the external auditory meatus of the speaker is picked up. The S/N ratio of a spoken-voice sound pickup signal is improved by further providing a noise-cancelling processing unit that reduces noise propagating in the inner space in which the microphone is arranged. Further, sound quality is improved by subjecting the sound pickup signal picked up by the microphone to an equalizing process for reducing muffled noise generated at the time of picking up spoken voice via the external auditory meatus.