We tested the sexual responsiveness of female canaries, Serinus canaria, to two sets of different types of female conspecific songs versus an adult male conspecific song. Female songs were either spontaneously emitted (‘S-songs’) or were testosterone-induced (‘T-songs’). Copulation-solicitation displays (CSD) were used as an index of female sexual response. Playbacks were performed several days before and during egg laying, a period of natural sexual responsiveness of the females to song. We demonstrated that the weaker sexual displays of female canaries were recorded to S-songs, thus suggesting that these types of female songs do not contain fully functional sexual releasers. Three T-songs elicited high levels of sexual displays, thus demonstrating that testosterone treatment may induce sexual release quality in the female songs. Study of the phonology of these three T-songs strongly suggested that special song phrases may be good candidates as powerful sexual releasers. To test the sexual value of these female song phrases, we carried out a third experiment, using hybrid songs where each of these special T-song phrase types was included in a well-known heterospecific context. Two phrases elicited high levels of sexual responses in females. Essential features of the male full song, such as broadband rapid frequency modulations and high repetition rate, are retrieved in both female song phrases. Taken together, these data demonstrate that testosterone treatment not only induces a male-like structure in the songs of females, but also induces functionally ‘male-like’ songs. This result allows features of the vocal control network of testosterone-treated females to be compared with those of adult males singing full songs, to distinguish neural correlates of testosterone-dependent full songs. However, because testosterone does not induce functionally male-like songs in all the females, neuroanatomical structure-function correlations need detailed behavioural analysis.