Hi John,
As you said,I have noticed this problem.So,instead of getting the max over the samples in a block,I am tending to detect the RMS value of the samples in a block.Whether this way is available to fix this algorithm flaw?
As in the source code,lower-accuracy approximation to dB is used to do a quicker converting to dB value,which incurs a little bit effect in the rang lower than "-40dB".But I have no idea to convert the linear value to dB value.Could you lend me some examples or instruction?
Best Regards,
Liang
