This idea is also published in the SE portal as SE-I-1382.
Some people have a low frequency voice timber that is very difficult to hear when reproduced electronically. Injection of some sound processing logic (equalization) to lessen the low frequencies and boost the mid-range and high frequencies can make the most difficult to hear voice much clearer.
Based upon data from Bob Heil, a renowned sound engineer from Heil Sound and an acknowledged expert at long-distance radio communications, a simple three band parametric equalizer circuit can result in massive speech recognition improvements, particularly in noisy environments. These circuits can be built in software; the same software that we are already using for noise reduction. Software circuits even allow for the frequency center of the three bands to be software defined and customized. Please note that while a three-band parametric equalizer should be sufficient for voice enhancement, additional bands could be added if deemed desirable. There should also be a set of absolute high/low frequency limits, beyond which the nothing is allowed.
This form of software parametric equalizer exists and is built into high end radios used for extreme long-range terrestrial and space communications.
A reasonable set of initial default values for a three-band parametric equalizer would be:
Setting |
Value |
Absolute Low Frequency Limit |
200Hz |
Absolute High Frequency Limit |
3000Hz |
Low Band Center Frequency |
200Hz |
Mid Band Center Frequency |
900Hz |
High Band Center Frequency |
2400Hz |
Low Band Gain |
-10dB |
Mid Band Gain |
-6dB |
High Band Gain |
+8dB |
I would propose that the RC App and RCV include equalizer logic and personalized settings to allow low/mid/high band adjustments from -16dB to +16dB on the transmitted data stream. This setting should be retained for the user as a ‘sticky’ setting. This should have an easily accessible on/off switch.
A ‘nice’ extension would be the option to perform the same equalization on received audio; overall audio for the RC App and possibly per participant in RCV.
It would also be a wonderful feature if an AI based ‘auto-tuning’ option could be implemented!