A method for realistic, conversational signal-to-noise ratio estimation

The analysis of real-world conversational signal-to-noise ratios (SNRs) can provide insight into people's communicative strategies and difficulties guide the development hearing devices. However, measuring SNRs accurately is challenging in everyday recording conditions which only a mixture sound sources be captured. This study introduces method for accurate situ SNR estimation where speech signal target talker natural conversation captured by cheek-mounted microphone, adjusted free-field convolved with measured impulse response to estimate its power at receiving talker. A microphone near receiver provides noise-only component through voice activity detection. applied recordings conversations two scenarios. It shown that broadband level distributions are estimated more proposed compared typical single-channel method, especially challenging, low-SNR environments. application two-channel may render realistic estimates valuable input instrument processing whose operating points determined estimates.