A continuous phase modulation method comprises the following steps: receiving a sequence of digital data symbols a(n) to be emitted; transforming the sequence of symbols a(n) to be emitted into a transformed sequence of symbols b(n), each symbol b(n) of which is equal to the sum of a symbol a(n) to be emitted and of a corrective factor equal to a transformation Tf applied to a plurality of differences (a(n)-a(n−1)) between two consecutive symbols to be emitted, the transformation Tf applied being a combination c of at least two differences between two consecutive symbols of the sequence to be emitted, transformed by the application of a non-linear function f; filtering the sequence of transformed symbols b(n) with a shaping filter and modulating the filtered sequence with a phase modulator; said transformation Tf being defined so as to minimize interference between modulated symbols filtered by a receiving filter.