audio processing

[Sebastian]

Yes, that's another main problem.. the maths. I hope to get the ALE software to see if it is possible just to store some coefficients and not calculate them within the pic. However (I think), all parameters change (all 5) just by changing one value (for example the gain) of a filter.. so every coefficient has to calculated, not only one.

Hm... I don't mind to spend some time working with the TAS3002 as it seams to be a very interesting IC and still easier than a "real" DSP, but first I should know how difficult it is and whether it is possible with a pic at all... I have a free C-compiler, but it can only manage 16 bit variables.. so thats the first border.. difficult as a 4.20 data format is needed.


Sebastian

[mister_e]

sure it is possible with a PIC. If you plan to do the 24-32 bit maths.. i would suggest to use a 18F PIC and it's hardware multiplier. Have a look to Microchip, there's certainely some interesting app note and code template for those maths

TAS3002 is a really nice chip. Audio quality is respectable... well for that price it's excellent. For sure nothing beat the pure analog stuff but you'll never be able to fit all TAS feature in a respectable PCB size using only analog stuff

For purist like me digital sound processing is not an option.. sound quality is too much altered... but that's easy, portable and 99% of the people are satisfied Anyway, how many people 'round the world says that MP3/WMA sounds good? Not me... but portable.

[Sebastian]

Hey.. a new problem. My free compiler does not support any PIC18Fxx, only 12Fxxx and 16Fxxx.
http://www.musicdsp.org/files/Audio-EQ-Cookbook.txt
Do you know if these formulas work for the biquad filters (for example bandpass)?
This would be a start.. if I don't get the ALE program. It doesn't seem to be too hard for a little pic (16F876), at least I will give it a try.

Sebastian

[mister_e]

As i remind you can enter your own coefficient in ALE and plot a graph of.

look at that, i think it could work... untested
http://www.gennum.com/audio/hip/soft...uad_filter.htm

[mister_e]

About a free C compiler, you may try the Microchip C18 student version.
http://www.microchip.com/stellent/id...&part=SW006011

i hate the microchip compiler but it's working. In another hand if you buy it... it will by far cheaper than Hi-Tech C.

CCS do one too. I just had a look to their website... seems to grow since last time i went there.

[Sebastian]

Thank you, the Excel sheet gives me a first impression of the calculations.
TI answered me.. "If you are a student, as you mentioned, I will like to inform you that we do not support students here at Texas Instruments." blabla....

Before trying the Microship compiler I'm going to test the 16F876 for it's maths capabilities.

regards,
Sebastian

[Sebastian]

Hm, I've tried to calculate the coefficients for a low pass filter with these formulas http://www.musicdsp.org/files/Audio-EQ-Cookbook.txt

Im comparison to the excel sheet I had to set in the "center frequenzy" and not the -3db frequenzy. So when calculating with 2000Hz (2100 in the sheet) and Q=1, I got the following results (41000Hz sample rate) for a low pass filter:

formulas from the link:
b0 = 0.0256695991
b1 = 0.0513391983
b2 = 0.0256695991
a1 = -1.8973216034
a2 = 0.3674095060


Excel sheet:
b0 = 0,026936651
b1 = 0,053873303
b2 = 0,026936651
a1 = -1,343504439
a2 = 0,451251044

I think the values are quite similar, however, I don't know if it's correct..
The next step will be the pic calculation...

regards
Sebastian