Output at specific Decibels

Hello, I work at a Psychophysiology lab and for the purposes of a few of our experiments we need to output sounds at specific decibel levels. These sounds will be played at the rest phase, after the majority of the experiment is done but they are still critical to our data. It is very important that the output be precise so that when we look at stress levels through skin conductance, we can be accurate in reporting our findings without confounds.

I understand that dB is a measure of sound intensity and as such it may be highly dependent on the hardware. But is there a way to use Audacity or another program to first measure and set the hardware (we will be using headphones in a quiet lab, only sounds should be buzzing/humming from computers presenting input and recording data from electrodes) at a certain ideal output dB from where simply adjusting a variable in Audacity when creating the sound would gives us the desired dB level. Currently the range we desire would be between 15-105 dB. Even if the range can’t be achieved, simply being able to produce a sound at a known dB level consistently for every single subject (~1200) over multiple computers and multiple testing centers (one is in LA and other in Philly) would be achievement enough.

Thanks for your help and I am sorry if this is a question that has already been answered or is easily resolved. A thorough Google search yielded little to nothing of value and while I have decent expertise with computers, I have almost none with audio equipment.

You want to determine dBSPL Sound Pressure Level, not just dB. The easiest way to do with is with a sound level meter and the cheapest one of those is by Radio Shack.


There’s an analog one, too, that I favor because you can’t use this one to measure actual music.

I don’t know of a good way to measure SPL in headphones. Last time I had anything to do with this, we had calibrated headphones and they told you on the published sheet what the SPL was at any applied power.

Once you establish that relationship, then you can use Audacity to increase and decrease accurately from there.

There’s no good way to start cold with Audacity and have it tell you dBSPL. It’s too open ended.


A quick note about dB. The dBSPL thing is a specific volume. It’s a specific amount of air moving in an acoustical manner (wind is not sound). A certain dBSPL value measured in Vienna with a Philips meter is exactly the same loudness/movement of air in New York measured with a Radio Shack meter. The only significant variation is the curves (A, B, C, etc.) which have to do with ear perception, and those are fixed and known, too.

A-Curve, 125dBSPL is a jet taking off at 100M. Doesn’t matter where on earth you and the jet are.

The dB inside Audacity has a digital reference not physical. 0dB is the place where the digital sound channel runs out of numbers. You can’t get any louder than that. All volumes are lower. -20dB in the US and -18dB in Europe is the tone at the beginning of a television show so the station can adjust the playback machines. It has nothing to do with the volume of sound in the room.

If you have brief access to an SPL meter, you can calibrate one particular microphone, preamplifier, amplifier, digitizer combination and use that over and over again to measure your own processes, but you can’t change anything. No volume controls, always fresh batteries, etc. etc. You have to put your microphone and electronics in a box and only take it out to use it for a measurement or re-calibrate it.

And that still won’t tell you the volume inside the headphones.


Thanks, your reply was very helpful in getting a better understanding of the actual phenomenon.

I also found this posted suggestion on how to measure the SPL of a headphone, do you think that would work?: http://demo.seanmulholland.com/blog/measuring-the-spl-volume-db-of/

Considering that this is an ancillary part of our testing, my lab manager just informed me that the new goal is to find a dBSPL (perhaps 85) and ensure that the headphones are set to that level consistently for all our subjects. Would this method of calibration be precise enough? And furthermore, how often would we need to do it? (Testing phase is over a year long and probably 4-6 per day, twins will be back to back as well). Ideally we wouldn’t want to do it more than every month.

Otherwise, could you suggest a place to purchase calibrated headphones.

Again, I really appreciate your help. Thanks.

It’s a good as any other, and it has the advantage that once you have the meter, you can repeat the test whenever you wish. I don’t expect the readings to “drift.” Once you get a reading with a standard amplifier at a standard volume control setting, it should stay that way until you change it.

You may find that at about the sixth or seventh headset, the changes and errors aren’t all that great. Once a company starts cranking out headphones, they use the same bins of parts over and over.

Component drift works in your favor here. 6 dB is a noticeable, but not very strong change in perceived volume – either direction, up or down. You can tell something happened, but it doesn’t affect the show very much.

In electronics values, however, that’s half and double the signal.

That’s why when people go shopping to upgrade their music system, going from a 30w amplifier to a 40w amplifier is a complete waste of time. You’d need to go to at least a 60w amplifier to hear any significant difference.

Measuring headphone volume is a little magic because unlike speakers, which you can measure in special rooms, headphones are pretty much tied to your ear canal. Anything you measure without the ear canal is a guess. I like the idea of using the weight to stabilize the measuring platform. That was a good idea.


Ah, I’ll try that then. Thanks a lot for your help, it was greatly appreciated.