Based on the knowledge and experiences from existing image steganalysis techniques, the overall objective of the paper is to evaluate existing audio steganography with a special focus on attacks in ad-hoc end-to-end media communications on the example of Voice over IP (VoIP) scenarios. One aspect is to understand operational requirements of recent steganographic techniques for VoIP applications. The other aspect is to elaborate possible steganalysis approaches applied to speech data. In particular we have examined existing VoIP applications with respect to their extensibility to steganographic algorithms. We have also paid attention to the part of steganalysis in PCM audio data which allows us to detect hidden communication while a running VoIP communication with the usage of the PCM codec. In our impelementation we use Jori's Voice over IP library by Jori Liesenborgs (JVOIPLIB) that provides primitives for a voice over IP communication. Finally we show first results of our prototypic implementation which extents the common VoIP scenario by the new feature of steganography. We also show the results for our PCM steganalyzer framework that is able to detect this kind of hidden communication by using a set of 13 first and second order statistics.