In order to solve the security problem of speech perception hash authentication, the application scope of speech authentication algorithm, and improve the robustness, discrimination and real-time authentication in the process of authentication, a multi-format speech BioHashing algorithm based on spectrogram is proposed. Firstly, the speech signal to be processed is converted into spectrogram and feature extraction is carried out by two-dimensional discrete cosine transform. Then, the dimensionality of the eigenvector is reduced by non-negative matrix factorization, and generation of BioHashing sequences by inner product of reduced dimension eigenvectors and orthogonal normalized random matrices. Finally, the BioHashing is encrypted by equal-length scrambling using Henon chaotic map. The algorithm also validates the unidirectionality of BioHashing with trapdoor by comparative difference method. The experimental results show that the proposed algorithm has the characteristics of good security, strong robustness, high real-time performance and wide application range. © 2020, Springer Science+Business Media, LLC, part of Springer Nature.