Safe Listening Levels

One of my biggest concerns as an audiologist for the open source project is the need to ensure SAFE LISTENING LEVELS. How this concern is addressed will likely be requested by most Investigational Review Boards. Clinically, I have seen more than one amateur audio engineer report an incident years before coming to the clinic of intense noise exposure while experimenting with sound. Sure enough, these individuals presented with a classic ‘noise notch’ in their audiogram.

It is always good practice when playing with sounds of unknown levels to slowly bring the earphones to your ears to be sure that whatever you are listening to will be at a comfortable/safe listening level. For those of us with normal hearing, this is one concern. When programming a device for someone with hearing loss, this is another concern altogether since there is often a need to make the sound output louder than normal-hearing listeners would tolerate. Therefore, a ‘listening check’ will not suffice. If this issue is ignored, there may be SUBSTANTIAL RISKS for the research participants. Furthermore, as the sound output is coupled deeper in the ear canal or if the ear canal is un-occluded (vs. supra-aural or circumaural headphones), it becomes more difficult to predict how individual ears will shape the sound across frequency. The only way to know this is to obtain probe microphone measurements near the tympanic membrane. These measurements will need to be compared to the measured or estimated (via scientifically-accepted prescriptive methods) loudness discomfort levels.

The Tympan unit comes with a potentiometer (blue volume knob) that controls gain. At first, I was concerned that this would introduce error in setting gain due to lack of precision in its placement. However, I learned that the ‘Serial Monitor’ in the Arduino software allows one to see the gain as the volume knob is turned (see figure). I assume that the adjustments are made in discrete steps of 5 dB and that it NOT the case that adjustments are continuous while the reporting of the gain is discrete. Another concern is that an experimenter or research participant will accidentally or intentionally turn the volume knob after it is set for the experiment. Therefore, it would be nice to be able to deactivate the volume knob after the gain is set or to conceal it (perhaps, something as simple as black electrical tape would suffice). Upon review of the example Tympan Library codes, it appears as if the former is easily accomplished with programming.

1 Like

This is great feedback. Yes, managing safe listening levels is indeed critical.

To confirm your observations:

  • Yes, the knob is completely controlled by one’s software. So, if you don’t want the knob, you can disable the knob by commenting out that line of the code.
  • Yes, in the default software that you received, the response of the knob is quantized into 5 dB steps. More or less quantization, along with more or less overall range, is easily adjusted depending upon your goals.

To provide the experimenter total control over the presented volume (and to prevent the test subject from having any opportunity to mistakenly adjust the presented volume), I agree that it would be good to defeat the volume knob.

Instead of totally defeating the knob, however, I might suggest that its programming be adjusted so that turning it all the way down will cut the sound to the earphones. This will give the subject an important sense that they can stop the sound at any time.

Finally, I find that the knob is most useful to the algorithm developer. While the developer is experimenting with their own algorithm, the knob can be set to change volume or, even more fun, any other parameter (or combination of parameters) that you’d like. Want the knob to control the gain just in the high frequencies? Do it! Want the knob to control the compression ratio? Do it! Want the knob to control pitch shifting? Do it! Interacting with your algorithm’s parameters in real-time can be very illuminating and can give great ideas on how to make them even better.

Chip

1 Like