Using Performance of ASR to Compute Optimal Location of Microphone

It has been observed that the measurement error in the microphone position from a fixed source location affected the room impulse response (RIR). This in turn affects the single-channel close microphone and multi-channel distant microphone speech recognition. Toward this end, we systematically study to identify the optimal location of the microphone, given an approximate and hence erroneous location of the microphone in 3D space. The primary idea is to use Monte-Carlo technique to generate a large number of random microphone positions around the erroneous microphone position and select the microphone position that results in the best performance of a general purpose automatic speech recognition (ASR). We experiment with clean and noisy speech and show that the optimal location of the microphone that achieves the best ASR performance is not only affected by noise characteristics but is also dependent on the SNR of the noise.