This paper describes the participation of the LIA laboratory to the Human Assisted Speaker Recognition (HASR) evaluation, which is part of the NIST-SRE 2010 campaign. The submission of the LIA for this task is based on a human decision. Samples were rated by three listeners, system decision being based on majority voting. Confidence scores were defined by mapping human decision to scores distribution of a SVM-based automatic system. This paper describes in section 3, the automatic system used for scores mapping is presented. In section 2 the algorithms used for listening stimuli generation and the protocol for samples listening and rating. Subsections 2.1 and 2.2 describe the algorithms used for automatic extracts selection from each model or test segment, and for extracts normalisation and concatenation. Subsection 2.3 describes the listeners involved and the listening protocol. Subsection 2.4 presents the calculations made on human decisions to obtain scores submitted to NIST. Finally, the characteristics of the submitted system are summarized and perspectives for future work are presented in section 5.