Question was something like, "What do you think of decibel meter X compared to decibel meter Y?"
Second question was something like, "Why is your method different/better than using a decibel meter?"
This is the long answer. You will need to understand what I did in THIS THREAD to follow the logic here.
The first point we need to make here is that YOU CAN NOT TRUST YOUR EARS. The response curve of the human ear is non-linear with respect to frequency. This means that you can hear two different frequencies which are exactly the same average power level and they will NOT sound like they are equally loud. You literally can not trust your ears. That said, if you trust your ears you are likely to come to a reasonable approximation of what "Billy Bob next door" does and does not hear. That is good enough for anyone who is only concerned about what his neighbors hear because their ears are as bad as your ears anyway. It is NOT good enough if you are wanting to prevent game from detecting your presence because game animals all have a different "hearing curve" from each other and from humans. Here are some interesting links about that. Dogs Hear, Cats Hear, Rabbits Hear, Squirrels Hear, Deer Hear, Pigeons are the most interesting of the lot...
Well clearly we are barely scratching the surface with recording equipment which is only intended to capture that which mere humans can hear. So there is your first take away, there is no simple way to study this problem scientifically. The only way to get absolute truth is to spend A LOT of money building a facility dedicated to the purpose and populated by equipment costing far more than the average person makes in a year. So we can either throw our hands up in disgust and declare that you can't measure it and you can't hear it so why bother, or we can endeavor to persevere
? Fair enough. One man's opinion is as good as another's and you can certainly do it that way, it just can't be duplicated but it has "sort of" worked for a long time.
Having detailed the futility of this work we can now blissfully ignore reality and declare that we ONLY CARE ABOUT HUMANS 
... problem solved. We don't care that deer, even squirrels can hear frequencies we can't even measure without a lab full of equipment. Now we can get to our trivial undertaking.
Lets talk about sound power. Specifically lets talk about RMS power level. RMS stands for Root Mean Square, it is the average power level over a finite time period. Maybe you want to know how much noise exposure you are getting from an airport. That is a continuous noise, perhaps it is busier during the day and less busy at night, but you can hear it continuously at your house. The average of that noise level (the RMS power) will vary based upon the "window" you look at when you average the sound. If you select a five minute window at 9am when the airport is very busy you might get an average level of 80 dBs RMS. That's pretty loud. If you select a five minute window at 11pm you might get 68 dBs because no plane landed during your five minute window. By the same token, you might use a one hour window and find that the RMS power at 9am is only 74 dB because you are averaging in more "quiet" time than you did with your 5 minute window. This same idea applies to our problem but in a very compressed way.
We want to measure a very high intensity, very short duration noise. If our "window" is too large, or our noise is not centered in the window we will get a different RMS power reading each time we measure it. If our window is too small and sampling doesn't index to the same start point we will get a different RMS power reading each time also. So before we can measure accurately we need to know when to start measuring and when to stop measuring. Enter the lowly "decibel meter" from Amazon. Most of them have a "slow" and a "fast" window. The less expensive ones have a "fast" window which is 125ms (that is 1/8 second). The duration of the event we are interested in measuring is about 1/10 (one tenth) of that. That means with the decibel meter we are averaging in 9 times more "data" than we want and it is "data" which is not "signal" but "noise". We only want about half the data contained in the "Just this part" shown on the inset (top left) below. The rest is noise. That entire sample is about 1/5 (one fifth) of the window on a low end "decibel meter" in fast mode.
The windowing function on a decibel meter works like this. It samples (at some unknown rate suppose 40kHz for convenience). As it samples it checks the absolute power level of the pulse it collects at that moment. It records that pulse and stores it in a bin which has a finite number of samples. For a 125ms window at 40kHz it would hold 5000 samples. Each time is samples it adds that data to the bin until the bin is full. (Remember this is a meter we can afford) At this point it checks the RMS value for that "window" and if it is larger than any measured thus far it displays it. It then empties the window and starts again checking every eighth of a second to see if that period was louder than the last period. A more expensive meter would have a "sliding window" and a very fast very expensive meter would run the RMS averaging process for every single sample. We are not talking about those measuring devices.
So the difference between recording the audio and analyzing the signal to identify the portion you want to measure and then isolating that portion and calculating the RMS power for that specific time period and using a $50.00 decibel meter to give you a "peak" value over a 125ms window is like the difference between a man with eyesight searching a room for a golf ball and a blind man searching the same room for that ball. Both men can do the job but one gets it done more quickly and with a whole lot less searching. Yes you CAN do it with a decibel meter.
So lets talk about that a moment. If you want to do it with a decibel meter, you have to take a LOT of samples, average them and take the standard deviation. That is you need to look at the collected data statistically. You do this already with muzzle velocities and chronographs and they are vastly more accurate than a dB meter. If you want to be fairly confident in your conclusions, I suggest you take a minimum of ten samples with each moderator then calculate the standard deviation and ES for that data in the same way that you do for muzzle velocities. You would do better to take thirty samples but ten is atleast more reliable than three or four samples averaged.
HOPE THAT HELPS
Second question was something like, "Why is your method different/better than using a decibel meter?"
This is the long answer. You will need to understand what I did in THIS THREAD to follow the logic here.
The first point we need to make here is that YOU CAN NOT TRUST YOUR EARS. The response curve of the human ear is non-linear with respect to frequency. This means that you can hear two different frequencies which are exactly the same average power level and they will NOT sound like they are equally loud. You literally can not trust your ears. That said, if you trust your ears you are likely to come to a reasonable approximation of what "Billy Bob next door" does and does not hear. That is good enough for anyone who is only concerned about what his neighbors hear because their ears are as bad as your ears anyway. It is NOT good enough if you are wanting to prevent game from detecting your presence because game animals all have a different "hearing curve" from each other and from humans. Here are some interesting links about that. Dogs Hear, Cats Hear, Rabbits Hear, Squirrels Hear, Deer Hear, Pigeons are the most interesting of the lot...
Well clearly we are barely scratching the surface with recording equipment which is only intended to capture that which mere humans can hear. So there is your first take away, there is no simple way to study this problem scientifically. The only way to get absolute truth is to spend A LOT of money building a facility dedicated to the purpose and populated by equipment costing far more than the average person makes in a year. So we can either throw our hands up in disgust and declare that you can't measure it and you can't hear it so why bother, or we can endeavor to persevere
? Fair enough. One man's opinion is as good as another's and you can certainly do it that way, it just can't be duplicated but it has "sort of" worked for a long time.Having detailed the futility of this work we can now blissfully ignore reality and declare that we ONLY CARE ABOUT HUMANS
... problem solved. We don't care that deer, even squirrels can hear frequencies we can't even measure without a lab full of equipment. Now we can get to our trivial undertaking.Lets talk about sound power. Specifically lets talk about RMS power level. RMS stands for Root Mean Square, it is the average power level over a finite time period. Maybe you want to know how much noise exposure you are getting from an airport. That is a continuous noise, perhaps it is busier during the day and less busy at night, but you can hear it continuously at your house. The average of that noise level (the RMS power) will vary based upon the "window" you look at when you average the sound. If you select a five minute window at 9am when the airport is very busy you might get an average level of 80 dBs RMS. That's pretty loud. If you select a five minute window at 11pm you might get 68 dBs because no plane landed during your five minute window. By the same token, you might use a one hour window and find that the RMS power at 9am is only 74 dB because you are averaging in more "quiet" time than you did with your 5 minute window. This same idea applies to our problem but in a very compressed way.
We want to measure a very high intensity, very short duration noise. If our "window" is too large, or our noise is not centered in the window we will get a different RMS power reading each time we measure it. If our window is too small and sampling doesn't index to the same start point we will get a different RMS power reading each time also. So before we can measure accurately we need to know when to start measuring and when to stop measuring. Enter the lowly "decibel meter" from Amazon. Most of them have a "slow" and a "fast" window. The less expensive ones have a "fast" window which is 125ms (that is 1/8 second). The duration of the event we are interested in measuring is about 1/10 (one tenth) of that. That means with the decibel meter we are averaging in 9 times more "data" than we want and it is "data" which is not "signal" but "noise". We only want about half the data contained in the "Just this part" shown on the inset (top left) below. The rest is noise. That entire sample is about 1/5 (one fifth) of the window on a low end "decibel meter" in fast mode.
The windowing function on a decibel meter works like this. It samples (at some unknown rate suppose 40kHz for convenience). As it samples it checks the absolute power level of the pulse it collects at that moment. It records that pulse and stores it in a bin which has a finite number of samples. For a 125ms window at 40kHz it would hold 5000 samples. Each time is samples it adds that data to the bin until the bin is full. (Remember this is a meter we can afford) At this point it checks the RMS value for that "window" and if it is larger than any measured thus far it displays it. It then empties the window and starts again checking every eighth of a second to see if that period was louder than the last period. A more expensive meter would have a "sliding window" and a very fast very expensive meter would run the RMS averaging process for every single sample. We are not talking about those measuring devices.
So the difference between recording the audio and analyzing the signal to identify the portion you want to measure and then isolating that portion and calculating the RMS power for that specific time period and using a $50.00 decibel meter to give you a "peak" value over a 125ms window is like the difference between a man with eyesight searching a room for a golf ball and a blind man searching the same room for that ball. Both men can do the job but one gets it done more quickly and with a whole lot less searching. Yes you CAN do it with a decibel meter.
So lets talk about that a moment. If you want to do it with a decibel meter, you have to take a LOT of samples, average them and take the standard deviation. That is you need to look at the collected data statistically. You do this already with muzzle velocities and chronographs and they are vastly more accurate than a dB meter. If you want to be fairly confident in your conclusions, I suggest you take a minimum of ten samples with each moderator then calculate the standard deviation and ES for that data in the same way that you do for muzzle velocities. You would do better to take thirty samples but ten is atleast more reliable than three or four samples averaged.
HOPE THAT HELPS
Last edited:
