Go Down

Topic: Removing high frequencies before digitizing audio. (Read 464 times) previous topic - next topic

CosmickGold

Oct 18, 2015, 01:26 am Last Edit: Oct 18, 2015, 05:56 pm by CosmickGold
I'm sampling at about 19k cps. Which means the highest frequency I can decently record is 9,500.

I'm having trouble getting rid of the higher frequencies without muffling the sound of the mid-frequencies too much. Right now, I've got muffled sound and high-frequency-distortion at the same time.

Has anyone designed the perfect low-pass filter circuit to solve this problem?

DVDdoug

I don't know how these things work, but there are switched capacitor filter chips.

I've owned a copy of Active Filter Cookbook, by Don Lancaster for a million years, but I don't think I've ever used it.   It shows you how to make filters using op-amps.   I'm sure you can find the same information online.

Quote
I'm having trouble getting rid of the higher frequencies without muffling the sound of the mid-frequencies too much. Right now, I've got muffled sound and high-frequency-distortion at the same time.
Have you tried a 1kHz or 5kHz sine wave, or something to make sure there's not something else going on (besides aliasing)?    Aliasing is caused  by high frequencies, but the results are not necessarily high frequency.  

Grumpy_Mike

Quote
Has anyone designed the perfect low-pass filter circuit to solve this problem?
Yes.
But is that your problem?
You do not get high frequency distortion with aliasing but you get random noises of any frequency.

What you need is a higher order filter, what have you got now? A sixth order filter should be good enough.

CosmickGold

#3
Oct 18, 2015, 05:55 pm Last Edit: Oct 18, 2015, 09:27 pm by CosmickGold
DVDdoug, Grumpy_Mike, your answers are both really good. But at this point, I withdraw my question.

Months ago, I tried two makes of .mp3 record/play breakout boards, and both were defective in design. That's what prompted me to work hard at getting the AVR to do all the work itself. I've spent most of my time these last two months in that attempt.

So many compromises between file size, bit rate, frequency range, filtering, noise immunity, gain control, as well as other AVR processes in competition with the audio (such as displaying seconds of play or record progress on the LCD screen).

At this point, I consider having the AVR handle all to be a lost cause.  The final blow being that saving a recording to the SD Card takes more time (when 1. checking the sector, 2. saving, 3. preparing the next sector) -- even on an Atmel1284p -- than there is on-chip memory for! In general, the sound-quality I can achieve is simply not good enough for me to sell when my project turns product.

So today, I have another breakout board I'm going to try, the FN-RM01, hoping it can do everything.

Again, thanks for your excellent answers.

Go Up