Detection threshold of uncorrelated (measurement) noise in HRTFs
* Presenting author
The present study investigates the influence of uncorrelated (measurement) noise added to a HRTF recording. Therefore, dummy head recordings of the Kemar Artifical Head Type 45BB where used and uncorrelated white noise added. The temporal and spectral effects are examined along with the underlying binaural processes that could make the noise audible. Furthermore, the detection thresholds for uncorrelated white noise added to the HRTF were measured for three HRTF lengths (256, 400 and 1024 samples at 44100 kHz sampling rate) and four different directions with pink noise bursts. The results are given over the SNR, computed as a ratio of the RMS of the first 256 samples (5.8 ms) of the impulse response (representing the direct sound of the HRTF) and the RMS of the noise. We found detection thresholds to be at around 25 dB SNR for the 256 and 400 sample length, increasing to around 32 dB SNR for a length of 1024 samples. These results are useful in designing parameters for HRTF recordings, especially for individualized HRTF recordings, since SNR is a factor of measurement time.