Design and evaluation of a microphone signal conditioning system

In voice recording applications, speech quality is frequently degraded by varying speech levels and microphone artifacts. In an earlier paper H. Brouckxon, K. Struyve and W. Verhelst (2004), we proposed three versions of an integrated signal conditioning system that joins the reduction of low-frequency microphone artefacts, known as `pops', to an existing dynamics processing architecture for level stabilisation. In the current paper, two formal subjective preference tests are described that have been used to compare and evaluate these signal conditioning systems. The tests show a general preference for two of our three proposed versions. By combining this result with computational complexity considerations, one preferred version for real-time implementation is proposed