While shopping at a grocery store recently, I noted another customer who was speaking continuously on their cell phone as they shopped. They were wearing a mask. They were speaking quite loudly. As this customer came near me, I noticed that his mask was being significantly "inflated" by his speaking and I observed that the air was blowing the mask away from his face and air was leaking in the gaps around the edges. It seemed to be a risk factor in failure of the mask in preventing transmission of Covid-19 from the wearer to others, including me.
I had previously read articles about transmission of droplets and mist during normal conversation. See the References section for some of these articles.
This project proposes a solution to help people to learn to speak less loudly on their cell phones. Reducing the volume of speaking should reduce the "blow by" air escaping masks. Therefore it ought to help reduce the spread of Covid-19. A side benefit to lowering people's volume when they are around others is that it reduces the noise disturbance that can be quite troubling.
Nature of This ProjectCell phone calls are handled by software applications that are used just for making calls. The Android phone calling application is available as open source software at https://android.googlesource.com/platform/packages/apps/Phone/+/refs/heads/eclair-sholes-release/src/com/android/phone. Even so, it is a very complex application and modifications to it are best left to those who are already familiar with the software.
For iOS phones, the calling application is proprietary and not available. Again, modifications are left to the experts.
As a result, this project will not actually implement the solution, but simply describe how it might be implemented. The solution is best left to the companies who make and maintain the phone calling apps used by cell phones. However, there is an opportunity for manufacturers to cooperate and provide a uniform solution that would be common across vendors. This would help socialize the new behavior of these phones, regardless of manufacturer. I will share more details on how these copies could cooperate a bit later.
Digitizing Sound for Cell Phone Audio TransmissionCell phone calling is performed by digitizing the input received by the microphone and thereby sending it as a digital signal. Digitization involves measuring the signal level at periodic sampling times and then representing that signal level as an 8 bit value, which means a number between 0 and 255. This means that only 256 different values have to be used to represent the softest and loudest sounds. We speak of 8-bits being the dynamic range. This is a fairly narrow range.
It was quickly discovered that a pre-amplifier must be used on the microphones of cell phones, because the human audio range is much greater than the available 8-bit dynamic range. Otherwise, the microphone could not both detect very large sounds and very soft sounds. We would need to speak with very narrow dynamic range or we would over-drive the microphone or under-drive it. However, with the pre-amplifier the phone adapts to the level of the sound it detects. If it is very large input sound, the amplifier makes it softer. If it is very quiet, the amplifier makes it louder. You can think of this pre-amplifier as translating all sound levels to be in a moderate range. The amplifier adapts over time, so you might start talking quietly and it needs to increase the volume, but then after a few minutes in the conversation, something agitates you and you start speaking louder. Then the pre-amplifier will reduce the volume.
Having such a variable pre-amplifier means that very high volumes get reduced to softer volumes. We could communicate just as well if we spoke softer, instead of at a high volume. Though we can speak at a wide dynamic range, it is fitted into a narrow dynamic range in order to fit into 8 bits.
This means that a person who speaks at a consistently high volume can communicate just as effectively as one speaking softer. And reducing the volume people use can help reduce the mask "blow-by" and lower the risk of transmission of Covid-19.
Proposed AlgorithmThe best scenario for implementing an algorithm to warn cell phone users against loud speaking is if the variable pre-amplifier communicates its settings in a way that is available to the phone calling application. If this capability is not available for certain models of phones, then a variation of the following algorithm can be used, though it will not be a flexible.
In summary, the algorithm will look for volume levels above a certain threshold. If that happens, then auditory and visual notifications are presented on the phone, briefly interrupting the call in progress. The point of these notifications and interruption is to cause the phone user to recognize they are speaking too loudly.
Above I mentioned that having commonality in the implementation of this algorithm among cell phone manufacturers would be the best. The volume threshold is one parameter that could be common. Another is the notifications. The notification sound could be constant. With land line phones, there was a common sound which was used to indicate the call recipient was already speaking on the phone. This "busy signal" was uniform and therefore it was recognized no matter which phone was being used from different locations. You can play some examples at https://en.wikipedia.org/wiki/Busy_signal. Probably a pulsing, at a slightly faster rate than a busy signal would work well for the "talking-too-loud" indicator. I will recommend the following specifically: two sine wave tones, one at 1000 Hz and the other at 1100 Hz. The 1000 Hz tone is first played for 200 ms, then 10 ms of silence, followed by 200 ms of the 1100 Hz tone, followed by 500 ms of silence. This sequence is repeated four times before stopping. Here is an example of such an audio notification: https://drive.google.com/open?id=15em012aoAG3Afgx84duA627gYtbsWw1N The visual notification should cover most, if not all of the phone's display, and be very plain. It should use just two colors: a white background and blue foreground. In the phone user's language it should say: "Don't speak so loudly".
For a threshold, the following recommendations are offered. Start the limit at 65 dB for 1 month, then decrease to 60 dB for the next month, 55 dB for the next month, finally settling at a threshold of 50 db for the fourth and following months.
As mentioned above, the implementation of the threshold detection is straight forward if the pre-amplifier gain setting is known to the phone calling application. If this is not possible due to phone construction, then an implementation where the pre-amplifier gain range is such that about 90% of full volume is where the threshold lies. Then, the detection of sound above 90% of full volume in any gain range is set as the threshold. Now, the 90% value is arbitrary, but the actual value should be chosen such that the pre-amplifier will automatically scale the the next lower gain range. Consider this example. A pre-amplifier has 10 distinct gain settings. When the volume level is below 80% of full value at a given gain setting, the gain is automatically lowered to the next lower gain range. Set the threshold at 81% of full value.
There may well be other types of pre-amplifier control systems, and the fallback approach can then be best determined by the phone manufacturer with the use of audio testing at the different threshold values of 65, 60, 55 and 50 db.
Implementation TermThis project does not need to have a permanent implementation, as it is primarily a protective measure against spread of the Covid-19 virus. However, the secondary effect of less noise disturbing others in public spaces might lean toward a more permanent implementation. I believe that after the 4 month ramp up, it will become easier for phone users to continue to use a softer voice.
Appeal for CooperationCell phone makers are fiercely competitive with each other. They try hard to differentiate themselves, making their products stand out from others.
As mentioned above, this project is a great opportunity for these makers to cooperate. They could implement this protection against Covid-19 in such a way that the results are uniform across all phones. I realize that this feature I propose could be very disruptive to some cell phone users. Of course, that is necessary for it to be effective in preventing loud speaking. I see a great advantage in the uniformity of the feature being that knowledge about it will spread more quickly among people who have phones from different manufacturers.
This project is an opportunity for the cell phone makers to provide a contribution to solving our health crisis. No doubt there will be goodwill toward them as a result. If they would all join together it would be an even more positive indicator to their customers and prospective customers.
References- https://www.sciencedirect.com/science/article/pii/S235208172030057X COVID 19 can spread through breathing, talking, study estimates. May 3, 2020. Current Medicine Research and Practice. Dr. Ramananda Ningthoujam, Ph.D.
- https://www.forbes.com/sites/brucelee/2020/04/15/speaking-alone-may-spread-covid-19-coronavirus-here-is-how-talking-sprays-droplets/#23030c392748 How You May Transmit COVID-19 Coronavirus From Talking Without Coughing Or Sneezing. April 15, 2020. Forbes. Bruce Y. Lee.
- http://blog.pnas.org/2020/04/fluid-dynamics-work-hints-at-whether-spoken-word-can-spread-covid-19/ Fluid dynamics work hints at whether spoken word can spread COVID-19. April 7, 2020. PNAS. Carolyn Beans.
- https://www.theverge.com/2020/4/17/21224815/talking-spit-watch-particles-mask-laser-droplets-visualization Video uses lasers to show how you spit when you talk. April 17, 2020. The Verge. Nicole Wetsman
Comments