The invention discloses a self-adaptive snore counteracting method and device based on personalized voiceprint training and a sleep earplug special for a patient, belongs to the technical field of intelligent
noise reduction and sleep assistance, and aims to solve the problem that sleep of the patient is disturbed by snoring in a ward. The method comprises a personalized training stage and a real-time recognition and counteracting stage, in the training stage, snore audios of snore patients in a ward are collected, acoustic features are extracted, and a personalized voiceprint model is established after
principal component analysis and
dimensionality reduction; in the real-time stage, target snore is accurately recognized after
dimensionality reduction is conducted through the same method, self-adaptive filtering is started to generate anti-phase sound
waves to achieve physical counteracting, and meanwhile key sound of a ward is reserved. The earplugs are made of medical-grade materials, are comfortable to wear, easy to clean and stable in endurance, accurately counteract snore without excessive
noise reduction hidden dangers, improve the
sleep quality of the patients, and are suitable for hospital wards,
rehabilitation centers and other scenes.