Gearslutz.com - View Single Post - Fuzzmeasure: a few quesions about frequency, reverberation and waterfall

JohnPM · 6th January 2010

I suspect you will need to spend some more time getting to grips with the measurement basics or your paper may end up with some howlers in it. I'm not sure how much you are already familiar with, so I'll try to outline the fundamentals.

The impulse response shows how an impulse emitted at the source position would be picked up at the measurement position. Some of that impulse is due to the direct sound from the source, altered by the characteristics and bandwidth of the speaker itself, some is due to the sound bouncing off the various surfaces of the room before arriving at the measurement position. The sound bouncing off the walls etc has to travel further to reach the mic, so it corresponds to later parts of the impulse. The frequency response you see is derived by doing an FFT of the portion of the impulse response you select - that way you can select a small portion near the beginning to see predominantly (or entirely, if short enough) the direct signal from the source, or a much wider portion to also see the room's contributions. The shorter the portion you select, however, the less frequency resolution you have and the higher the lowest valid frequency in the response.

A single RT60 figure for the whole response is not very meaningful, because the RT60 varies with frequency. It is more common (and useful) to calculate the RT60 figures at octave or one-third octave spacings, which is what FM is showing you. The RT60 figure is based on the time the response takes to decay by 60dB, but the "response" in that case is the envelope of the impulse response, not the impulse response itself. It is usually measured by doing a backwards integration of the impulse response and looking at the slope of the resulting line between defined levels - google "Schroeder Integration" and study ISO 3382 for the details of that.

You cannot determine RT60 times from waterfalls. The waterfall is produced by moving a window through the impulse response and plotting an FFT for each position, the apparent slope of levels on the waterfall will depend on the width of the window you are moving along the impulse.

Hope that helps.

6th January 2010	#3
JohnPM Gear nut Joined: May 2009 Posts: 141	I suspect you will need to spend some more time getting to grips with the measurement basics or your paper may end up with some howlers in it. I'm not sure how much you are already familiar with, so I'll try to outline the fundamentals. The impulse response shows how an impulse emitted at the source position would be picked up at the measurement position. Some of that impulse is due to the direct sound from the source, altered by the characteristics and bandwidth of the speaker itself, some is due to the sound bouncing off the various surfaces of the room before arriving at the measurement position. The sound bouncing off the walls etc has to travel further to reach the mic, so it corresponds to later parts of the impulse. The frequency response you see is derived by doing an FFT of the portion of the impulse response you select - that way you can select a small portion near the beginning to see predominantly (or entirely, if short enough) the direct signal from the source, or a much wider portion to also see the room's contributions. The shorter the portion you select, however, the less frequency resolution you have and the higher the lowest valid frequency in the response. A single RT60 figure for the whole response is not very meaningful, because the RT60 varies with frequency. It is more common (and useful) to calculate the RT60 figures at octave or one-third octave spacings, which is what FM is showing you. The RT60 figure is based on the time the response takes to decay by 60dB, but the "response" in that case is the envelope of the impulse response, not the impulse response itself. It is usually measured by doing a backwards integration of the impulse response and looking at the slope of the resulting line between defined levels - google "Schroeder Integration" and study ISO 3382 for the details of that. You cannot determine RT60 times from waterfalls. The waterfall is produced by moving a window through the impulse response and plotting an FFT for each position, the apparent slope of levels on the waterfall will depend on the width of the window you are moving along the impulse. Hope that helps. __________________ REW Author