Spam in Internet Telephony (SPIT) is a major threat for VOIP networks. To protect VOIP users, automated systems are required to identify and block spam generators (spitters). One of the modules for these systems is verifying whether a caller is human or a computer program. In this work, we present a method to distinguish between computer programs and human users. Simply stated, our method asks the user to repeat a sentence. The reply is analyzed to verify it is the requested sentence said by a human. The main advantage of our system is that it is entirely speech-based, making it more usable and accessible compared to today’s audio-CAPTCHA systems.