In this work, a humidity sensor capable of speech detection and recognition is proposed. In such a sensor, nanoforests with superhydrophilicity are integrated as the humidity sensitive material using a simple process compatible with conventional micromachining. Due to the chemisorption-physisorption of water molecules, and the capillary condensation of water molecules within the nanopores, sensitivity of the nanoforest-based humidity sensor can reach as large as 0.15%/%RH, while the response/recovery time for speech detection can be speeded up to 1.1/0.9 s. Such a humidity sensor is expected to help people with vocal cord injuries regain their ability to speak, besides, it is expected to be used for communication in special situations, and to assist in voiceprint anti-counterfeiting and recognition.