The invention discloses a modulation and recognition device and a method for union high-order statistic and spectral peak features. The modulation and recognition device comprises a signal preprocessing module, a high-order statistic feature extraction module, a spectral peak feature extraction module and a union recognition module. Signals to be recognized and modulated are input into the signal preprocessing module so that filtering, carrier frequency estimating and normalization processing can be carried out to obtain the preprocessing signals, and then the preprocessing signals are input into the high-order statistic feature extraction module and the spectral peak feature extraction module so that the features can be extracted, wherein the features include high-order moment features, high-order accumulation features, constellation cluster point values, feature power spectrum variance features, first-order differential amplitude histogram spectral peak number features and the like. The extracted feature information is input into the union recognition module. A classifier based on the union features carries out mode feature matching on the input signals and outputs recognition results. According to the device and the method, recognition of the SC-FDMA modulation mode is achieved for the first time, the recognition rate of the high-order QAM modulation mode is improved, the judgment threshold of recognition in MFSK classes is expanded, and the device and the method can be applied to the fields of frequency spectrum management, electronic countermeasures and the like.