To derive these estimates, signal and noise trials are denoted by S and N, respectively, to which individual (i ) counts of hits (H i ) and false alarms (F i ) are derived and, subsequently, their rates (hit rate = h i and false alarm rate = f i ).
Basically, this analysis uses false memory rates, hit rates and false alarm rates as the primary dependent variables, which were corrected prior to analysis as recommended by Snodgrass and Corwin [25] to account for deviations from the normal distribution due to positive skewness.
The main performance metrics considered in this evaluation were hit rate and byte hit rate.
The speech hit rate and non-speech hit rate are calculated as follows: (10).
3) The LTSD-VAD algorithm yields the best speech hit rate while non-speech hit rate is poor.
Hit rate, false alarm rate, sensitivity, and response bias are summarized in Table 2.
2) Harmfreq-VAD, LTSV-VAD, and LSFM-VAD algorithms also obtain a moderate behavior with relatively high non-speech hit rate but slightly low speech hit rate.
The experimental results show that the proposed method outperforms the other up-to-date VAD algorithms under the most non-stationary noisy environments and is more robust than other VAD algorithms even at low SNR due to the highest non-speech hit rate and a moderate speech hit rate.
In this case, throughput is proportional to hit rate, because the higher the hit rate, the more concurrent requests the cache is able to serve, whereas in the sequential execution of the simulations, throughput was dependent on the algorithmic overhead of the replacement strategy.
Sohn-VAD algorithm yields a moderate behavior with relatively high speech hit rate but slightly low non-speech hit rate.
