An Effective Test

I have found an effective method of performing pure-tone audiometry at ordinary loudness levels, used for tuning up a Crescendo.

It consists of listening to a monaural signal sent equally to left and right channels. The signal is a sinewave logarithmic sweep from 250 Hz to 10 kHz. I use two levels of the sweep, -23 dBFS and -43 dBFS. The reason for these particular levels will be discussed below.

But first, the rationale for this test is that a properly set up Crescendo should allow you to hear the full sweep, perfectly centered in your mind’s eye, neither deviating to the left nor right as the sweep progresses. And it ought to sound roughly constant in loudness.

Here is my test oscillator setup for the sweep:

The DIM button allows for an immediate -20 dB cut in output level, giving me my lower amplitude test with a single button click.

This sinesweep is then fed through an equalizer set up to approximate constant Phon loudness across the frequencies of the sweep. My attempt at this is shown here:

(Heh! You notice that the use of a logarithmic frequency display exaggerates the importance of the bass region, taking almost half the display for frequencies below 500 Hz. In Bark frequency space this same region occupies 1/5 of the frequency span.)

The important thing here are the frequencies of the filters, their boost/cut amounts, and their Q values. This is not quite correct, but close enough for our purposes.

Now using this filtered sweep alone, into Crescendo, has Crescendo idling at no compensation until a bit of signal creeps into one of its processing channels. And then Crescendo goes from nothing to full-bore compensation. This can cause noticeable jumps in loudness as the sweep progresses.

To overcome that artifact, we use a parallel channel in the mixer to send along some low level pink noise. Not enough to interfere with the test, but sufficient to keep all the Crescendo channels opened up and operating.

Now that we have Crescendo’s full attention, I use the -23 dBFS level setting for the sine sweep because, in my lab, I have my Crescendo calibrated so that a -23 dBFS amplitude sinewave, which measures -23 LUFS in loudness meters, produces a signal at the output of my headphones of 73 dBC. So my Crescendo is set up with Cal dBFS = -23 and Cal dBC = 73. This actually produces an input signal of 70 Phon into Crescendo. Your settings will vary, depending on your chosen calibration settings.

What I’m trying to establish here are two sinesweep signal levels that produce 70 Phon and 50 Phon into Crescendo. Both of these are within normal music listening range. And neither is an extreme setting.

For adjusting Crescendo I start with the louder sine sweep, and listen with my eyes closed. As the sweep progresses, with no corrections being produced from Crescendo, I hear the sweep starting in the middle, then veering to the right, then disappearing, and finally reappearing at its highest frequencies on the left side. We want to adjust Crescendo so that the signal remains close to the same loudness all the way up, and centered in the stereo field.

Getting the signal to remain at constant loudness depends on making elevation settings at each frequency. This is also how you get the signal centered. Don’t worry for now about +D/-H settings, just leave them at 0.

Most of us have our best hearing in the bass region, so you could take the 250 Hz start of sweep as your loudness reference. Raise the elevation settings at each other frequency so that the sweep is audible and approximately the same loudness.

Then for final touch-up, DIM the sweep and readjust the thresholds for centered constant loudness. At the low loudness levels, this is a more sensitive test for what your effective threshold elevations are.

Any disparity between what your settings are at the low signal level, compared to your original settings for the louder sweep, is an indication that you may have decruitment in addition to normal recruitment hearing.

  • DM

When I go through this exercise for myself, I find that the resulting audiology settings are much less severe than the audiology from my threshold level testing in an isolation booth. We don’t live in a threshold level world. Most of my time is spent monitoring sounds between 50-70 Phon.

Here is the result from my own adjustments:

The fact that my settings level out where they do indicates that I found no perceptible improvement by pushing the elevations any higher at the higher frequencies. The level value corresponds to the minimum necessary threshold elevation settings needed to achieve a satisfactory, constant level, stereo centered sine sweep.

When starting out, in order to get ballpark settings for threshold elevations, I found it helpful to feed just one channel, left or right, with the sweep. That avoids the confusion of two things going on at the same time. Once the rough settings were obtained that way, then feed the sweep to both channels for the centering exercise.

And by starting from scratch like this, I found no particular indication that I need any decruitment adjustments. I’ll live with these new settings for a while, and maybe tweak them as time goes on. But for now this sounds pretty darn good to me.