Have a question about multimedia or audio? Ask our Expert, Ken Boyce

Ken:   I wrote a lengthy response to your letter to me.  When I pushed "send" a message flashed saying I was unauthorized and then deleted my message.  That was very bad. Can you help rectify this.

Ken: This is once again your ancient engineer Al Mond, who celebrated big nine-0 birthday this week, asking your advice.  Instead of trying to create and build my idea of a perfect complementary audio hearing aid, let me put the question of you.    Do you suppose it is possible to build an audio amplifier which  deletes all vowels and amplifies only the consonants?  The technical difference is that the vowels are all sinusoidal and the consonants are literally noise spikes.   There is a huge theory that goes with this and I am hoping to publish this in the New England Journal of Medicine.  But the bottom line is that hard of hearing people typically don't hear the consonants.  For one simple but important reason, the consonants are thousands of times quieter than the vowels, and considerably shorter in duration.  Give me your thoughts of the possibility of a consonants only audio amplifier.

Hello Nate,

I hope your Thanksgiving was all it should be....with family and friends.  Also, CONGRATULATIONS on your 90th.  I hope I am still as inventive and interested as you are when I am 90.

Nate....your question regarding an amplifier that only amplifies consonants and rejects vowels is way out of my league.  My natural response would be to say that you would need very sophisticated signal identification and  filtering to accomplish the task.   I personally don't know any techniques in analog signal processing that could do that job economically or with a high degree of programmability and repeatability (stability).

I think you would have to resort to a digital signal processor programmed to recognize the audio type.

The minimum HW in the DSP scenario is an A-D converter, the DSP, and then a D-A.  The processing requirements would dictate the power consumption overall.  Because it is mono voice being processed, the sample rates can be kept low, but to recognize the sound type on the fly will still need many calculations on each sample and its relationship to previous samples.

I am sorry I am not much additional help to you at this point.   I know a couple of good DSP algorithm people, and next week I will pose the question to them and see what I get back.

Kind regards,
Ken

Hello Nate,

I hope your Thanksgiving was all it should be....with family and friends.  Also, CONGRATULATIONS on your 90th.  I hope I am still as inventive and interested as you are when I am 90.

Nate....your question regarding an amplifier that only amplifies consonants and rejects vowels is way out of my league.  My natural response would be to say that you would need very sophisticated signal identification and  filtering to accomplish the task.   I personally don't know any techniques in analog signal processing that could do that job economically or with a high degree of programmability and repeatability (stability).

I think you would have to resort to a digital signal processor programmed to recognize the audio type.

The minimum HW in the DSP scenario is an A-D converter, the DSP, and then a D-A.  The processing requirements would dictate the power consumption overall.  Because it is mono voice being processed, the sample rates can be kept low, but to recognize the sound type on the fly will still need many calculations on each sample and its relationship to previous samples.

I am sorry I am not much additional help to you at this point.   I know a couple of good DSP algorithm people, and next week I will pose the question to them and see what I get back.

Kind regards,
Ken