In the case of extracellular recordings, spike detection algorithms are necessary in order to retrieve information about neuronal activity form the data. We present a new spike detection algorithm which is based on methods from the field of blind equalization and beamforming. In contrast to existing approaches, our method estimates several waveforms directly from the data and corresponding linear filters are constructed. The estimation is done in an unsupervised manner, and the few parameters of the algorithm are intuitive to set. The algorithm allows for superior detection performance, even when multiple neurons with various waveforms are present in the data. We compare our method with current state-of-the-art spike detection algorithms, and show that the proposed method achieves favorable results.