Previous Testing Method



Measurement equipment includes Josephson Engineering free-field measurement microphone model C550F with 6mm 3.5um diaphragm positioned one meter left and level with muzzle, orientated toward the source, and 1.6 meters above dirt/grass. Digital samples are taken from balanced signal path by Tascam US122 at 10.4 μs intervals (96,000 per second) and 24 bit resolution (16.7 million possible steps). Recording hardware rise-time has been measured by an audio engineer at 20 microseconds while the microphone is estimated (using the 0.35/BW formula) to have a rise-time of 15.9 microseconds. Peak data interpretation and sound-pressure to sound-level translation by Sony Vegas software ensures single-event peak (wave crest) detection. Total system calibration during each use with Bruel & Kjaer model 4220 Pistonphone. Sound analysis is linear to better correlate with perception and to best record actual event data and therefore does not meet MIL-STD-1474D which specifies A weighing (designed to correlate with hearing loss rather than perception). Linear measurement results in higher levels than A weighing and this data should not be compared to MIL-STD data. Temperature and humidity measurements by Kestrel 3000. Silencers will be examined for evidence of artificial environments. If present, this will be noted unless the medium can be removed.



It is important to realize the limitations in this data for comparison to manufacturer's numbers or other published data. While this data is captured in good faith and without bias under somewhat controlled circumstances, it is not comparable to other environments or test equipment. An effort was made to have the data be reproducible and correlate with what a human observer would agree with, but your results may vary. This data must be interpreted with knowledge of the limitations of testing and how environmental factors such as temperature, humidity, and location effect the results. We do try to compare related products in the same session to make this data as comparable as possible, but if a manufacture's quoted data is different it does not mean that their data is any less valid. Individual units may vary depending on age, condition, manufacturing tolerances, and cleanliness.

Because the high-speed computerized metering system is calibrated during each use, the rise time exceeds military specifications, and the actual 'crest' of the highest wave is measured -- the results should be useful and comparable to other results under similar conditions.

Finally, there is more to a silencer that sound levels. Durability, weight, accuracy, length, attachment method, customer support, etc. all matter so please don't make a decision based on these numbers alone.


How we test:  

1. Install the US122 (Fig 1) and driver. Configure switches as seen in photo. Input L should be set to Mic/Line and phantom power should be on. Use the Mic-in L connector.

Fig 1

2. Plug the Microphone (Fig 2) into the US122 with an XLR cable and a 20 db attenuator (Fig 3). Mic should be on a mic-stand (and not on a table with the gun as vibrations move faster through dense materials than through air) and the same height as the firearms muzzle, 1.6 meters above the ground. Orient the mic toward the source (as this is a free-field microphone). Use a shooting fixture to hold the firearm muzzle in a precise location, and ensure that this is precisely 1.0 meters from the microphone. We have a mic-stand with a steel rod which ensures that the mic diaphragm is very close to 1.0 meters from the muzzle.

Fig 2 Fig 3 Fig 4

3. Set Input-L dial on US122 around the 6th line for a CAR-15 and around the 7th line for a .22LR. Other weapons might be somewhere in between. Using 24 bit recording software such as Audacity, record some 124.0 tone with the Pistonphone mic calibrator (Fig 4) at 96Khz and then pause the recording. Make sure to set the preferences to 24 bit recording and 'uncompressed export format' to '32-bit float.' Prepare the firearm. Continue recording, and then fire 10 shots suppressed at about 2 second intervals. Pause the recording. Unmount the silencer. Continue recording. Fire the gun 10 more shots. Stop the recording and save the file for analysis. Now go and examine the file for evidence of clipping. If the unsuppressed gunshot touches the edge of the range, the results are invalid and you must turn down the gain dial and try again. Also if the unsuppressed spike does not show much modulation, then turn up the gain and try again. You would like to be between 50% and 100% modulation for the unsuppressed shots. A reasonable file will look like Fig 5.  The middle image (Fig 6) is a close-up of the suppressed shot, and the final of the unsuppressed shot (Fig 7).  Notice how the spike and its decay (A-duration) in this unsuppressed shot lasts less than 1/1500 of a second. This means that reflections from the ground or ceiling cannot effect the peak results in this example unless, perhaps, they are within a 4 to 5 inches of the equipment. Because the A-duration in suppressed shots is shorter and less intense, reflections are more likely to play some form of role so it is best to stay away from other objects.


Fig 5 Fig 6 Fig 7


Now that you are done gathering data, we can analyze it:

Load the file into the free trial or full version of Sony Vegas video editing software. Right-click on the master-bus sound level meters and make sure 'hold-peaks' is on. Play the audio file previously recorded. As the tone is playing, note the level at the top of the meter (Fig 8). When the suppressed-shots occur, note the levels again for each shot (Fig 9). When the unsuppressed shots occur, note these 10 levels  (Fig 10). Now average them and add the difference of the levels to 124.0 dB to arrive at the peak gun-shot level.

Fig 8 Fig 9 Fig 10


An Excel spreadsheet that we use to do the math can be downloaded to interpret the results. This spreadsheet accepts Sony Vegas dBU values as input, not SPL numbers. The dBU levels can be entered either with or without the '-' sign as absolute values are used.

For RMS (not peak) spectral graphs (Fig 11) we use Spectra RTA software. RMS is the peak level integrated over time, and is useful to show the total sound 'amount' in each band (the 'area under the curve' so to say). You can also get fancy and overlay the two graphs (suppressed and unsuppressed) -- to see how much noise is 'removed' and from where in the audio band from the sound signature. We overlay them in Photoshop.

Configure Spectra RTA to 'RTA mode, left channel only' and set the spectrum analyzer to 1/24 octaves, 30Hz to 20Khz. Turn 'Averaging' to off and 'Peak hold' to 'Forever.' You may set the weighting to 'Flat' or 'A' as desired.

To calibrate the system, start playing the .wav file with Audacity or Vegas and hit "NEXT" on Spectra RTA's calibration procedure. While the calibration tone is playing, enter 124.0 into the 'Sound pressure level at the microphone' dialog. Hit 'Start.'

Now continue playing back the .wav with Audacity or Vegas while Spectra RTA is running. When you see Audacity's timeline approach the blast impulse, get ready to hit 'stop' on Spectra RTA's meter. Hit 'stop' when you see the graph jump. This will be 1/2 second or so after you hear the impulse. The meter will freeze.

To save a graph, select 'Copy image as bitmap' and then load an image-editing program, select new, then paste, then save this image.


While interpreting the graph, note that high frequencies do not travel as far, so while a high-pitched-sounding silencer might appear to be 'louder' to the shooter, it may be much more quiet to a distant observer. What may appear to be a high-frequency boost might really be a low-frequency cut. Depending on your goal, either may be more or less desirable. If your goal is comfortable shooting, then look for a high-frequency cut. If your goal is stealth, then perhaps a low-frequency cut is best.

Fig 11

Finally, these numbers are 'unweighted linear' and are not the same as the 'A' weighted results most manufactures publish. Some people believe that linear makes more sense for measuring human perception of gun shot noise but is problematic because it does not agree with archival data. Also note that a semi-auto is louder on the right side than the left because of ejection port noise, and that indicated dB SPL can fluctuate depending on test conditions. Don't get fixated on 0.1 or 0.5 dB differences as they are not significant. If one silencer is within 2 dB of another, then they are nearly equal in peak sound and one should look to other characteristics to decide which unit you prefer.

Our raw data files are available for download so that others may confirm the results and see what complete time/pressure traces look like for each silencer. They are in .wav format and viewable with sound editors such as Audacity, although they are not really useful to play as sounds because of the extreme dynamic range.


Contents are Copyright 2005 by Robert Silvers and may not be reproduced without permission.

Back to previous page