There are various challenges in speaker recognition mainly due to lack of speaker databases having speech data addressing the several issues affecting the efficiency of the speaker recognition systems like session variability, channel variability, with-in and between speaker variabilities. Due to scarcity of the speaker databases, it is intended to collect a database which can address these problems of speaker recognition systems. This paper describes a protocol developed for collecting database of speech samples in Indian scenario for further research work. First, the requirements and the rationale for these requirements are described. Then, description of the technical equipment and setup employed is given. Finally, the post-processing of the recordings is explained to prepare them for inclusion in the database.