Automatic speech recognition in a home environment is currently of large interest for many practical applications. It is also challenging for signal processing, since several problems have to be solved satisfactorily. First, a large degradation of speech recognition performance is caused by room reverberations. Second, environmental noises like a radio, television or kitchen devices may further degrade speech recognition. And third, the problem of concurrent and possibly moving speakers has to be addressed. In this paper we present our results of an extensive investigation of state-of-the-art multi-channel signal processing algorithms used for dereverberation, noise reduction and speaker localization. The investigation is based on a microphone array with 16 microphones which are part of a fixed ceiling-mounted device. All signal processing algorithms are implemented on a PC platform to run in real-time and produce an enhanced signal, which is finally used for speech recognition.