When I ran simulate_trajectory.py using wavfile.read , which read the data as int16, I convolved the int16 data with the rir calculated by gpurir and then I used wavfile.write to write the output data. And I found that datas are all overflowed. When I normalize the value of input to [-1,1], the output was small . My microphone distance is 2.5cm, source to microphone distance is 10cm.
Do I need to normalize the input to [-1,1] before convolve with rir?
When I ran simulate_trajectory.py using wavfile.read , which read the data as int16, I convolved the int16 data with the rir calculated by gpurir and then I used wavfile.write to write the output data. And I found that datas are all overflowed. When I normalize the value of input to [-1,1], the output was small . My microphone distance is 2.5cm, source to microphone distance is 10cm.
Do I need to normalize the input to [-1,1] before convolve with rir?