This article summarizes the reflexion carried out during the first phase of SAM-A Esprit project (819, on the topic of assessment methodology for speaker identification and verification systems (task 2500). We underline the need of standard evaluation protocols and we propose several ideas to reach this goal.