Abstract: To compensate for the poor signal-to-noise ratio when transforming a patient tracheal sound signal into a respiration flow signal, the transformed signal is used when the sound signal is above a threshold, and interpolated values are used when the sound signal is below the threshold. A volume of breath signal is reliably obtained. Respiration start/stop is detected either by analyzing the sound signal or by physical measurement of the patient. A wearable device incorporating the sound signal processing method can provide an immediate indication or alarm when the volume of breath signal indicates a condition requiring medical attention.