Time dependent frequency and dBA filtering

I am planning on making a program that will input a .wav file and filter out the signal so the output will be within the OSHA occupational noise exposure limits for my digital signal processing class. The issue I am having is the filtering has to do with how long a person is exposed. For example a person can only be exposed to 90 dBA for 8 hours a day and for every 5 dBA increase the time is cut in half so 95 dBA is 4 hours and so on. Is there any way to add up the total time a specific frequency or sound level is present and filter it if it goes above its designated time limit. Any help would be greatly appreciated.

Thank you

[Daniel Shub]

The wav file format does not specify sound pressure, but rather digital units which are converted to a voltage by your AD device (e.g., a soundcard) and eventually to a pressure waveform by your transducer. Are you asking how to calculate the maximum gain on your digital unit to sound pressure transfer function such that the resulting sound signal will be exactly at the OSHA limits? Are you going to allow a time dependent gain such that loud periods of time are amplified less/attenuated more?

[sodonnell]

Thank you for your response, basically I am just trying to find out the most efficient way to filter the audio signal so that it lies within OSHA's requirements.

Answer 1

[Walter Roberson]

Why not just filter to the dBA that can be listened to all day? Is there some "cost" associated with filtering that makes it worth allowing the high volumes through part of the time?

With regards to total time a specific frequency is present: is it definitely by frequency? Or is it effectively by total energy? For example, if it were by frequency then 4 hours at 95 dBA at 5 Hz could be followed by 4 hours at 95 dBA at 8 kHz, as neither one would exceed the limit for the frequency, but the total would be 8 hours at 95 dBA anyhow.

If the limit is by dBA not by frequency, then the limit is by (power * time): 95 dBA is twice the power as 90 dBA, and twice the power for half the time gives the same (power * time). The integral of power over time is "work", the SI unit for which is Joule. I think you should be able to calculate the total energy associated with an audio sample much more simply than tracking by frequency.

[sodonnell]

Thank you for the response. It is based on the dBA per hour in a workday. You can see the chart from this link https://www.osha.gov/pls/oshaweb/owadisp.show_document?p_table=standards&p_id=9735