Provided are a named-entity recognition method (200), a named-entity recognition apparatus (700), a named-entity recognition device (100), and a medium, said method comprising: acquiring a voice signal in a specific scenario (S201); extracting a voice feature vector in the voice signal (S202); extracting, on the basis of a text result after voice recognition of the voice signal, a text feature vector in the text result (S203); combining the voice feature vector and the text feature vector to obtain a composite feature vector of each character in the voice signal (S204); and processing the composite feature vector of each character in the voice signal by means of a depth learning model, so as to obtain a named-entity recognition result (S205).
展开▼