Blind Estimation of Room Acoustic Parameters and Speech Transmission Index using MTF-based CNNs

Room acoustic parameters, such as reverberation time <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">(T₆₀)</tex> and clarity <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">(C₈₀)</tex> , as well as the speech transmission index (STI), are essential in acoustics. However, such parameters and STI are difficult to obtain in everyday places where people exist. Blind estimation for those parameters without measuring room impulse response (RIR) is necessary and challenging. This paper proposes a method based on the modulation transfer function (MTF) and Schroeder's RIR model for estimating <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">T₆₀s</tex> in seven-octave bands. The estimated <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">T₆₀s</tex> are used to approximate the MTF and RIR. Consequently, the STI and five room-acoustic parameters, including <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">T₆₀</tex> , early decay time (EDT), <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">C₈₀</tex> , Deutlichkeit <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">(D₅₀)</tex> , and center time <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">(Tₛ)</tex> , can be estimated. We deploy convolutional neural networks for mapping temporal amplitude envelopes of a reverberated speech signal to <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">T₆₀s</tex> for the sub-bands. Simulations were carried out by estimating the five parameters and STI from unseen reverberated speech signals. The root-mean-square errors between ground-truths and estimated parameters suggest that the accuracy of the estimated <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">T₆₀</tex> and STIs can be improved by about 40% and 25% compared with previous methods, respectively. The other parameters were also correctly estimated, and they are comparable with those obtained from standard measurements.