best practice for hi-q ATM168 audio ADC

i'm curious know if there's i've missed altogether or perhaps if there's smarter way go when trying adc on board.
many people write it's possible, when balancing processing 
here's setup:

i have arduino duemilanove (atm168) far standardfirmata on board.
on breadboard have line in stereo, dc offset raise level wanted 0~1023 values.

for actual adc i'm aware actual adc circuit best way go, i'm trying arduino microcontroller. david mellis (arduino team) has created sensor library in order achieve higher sampling rate (44100) when using arduino microcontroller adc, i'm not quite sure yet how implement it. find code confusing whentrying understand how 'sync' data arduino program.

ultimately using openframeworks use data, i'm trying figure out how best achieve sufficiently resolutioned (fft) spectogram representation of analog audio signal.
i not using audio reproduction, rather make hash values ("fingerprints") use characteristics of audio realtime comparison source.

i'd know inputs in order best arduino pass digital audio data on serial computer.
do have best pratices hashing audio - perhaps on arduino?

i'm bit confused serial speed standard firmata - perhaps mirocontroller can handle adc/ data, can serial port keep speed? why data seem low resolution?

as can see there few issues i'm juggling here , i'm @ standstill being beginner .. code best uses potential of microcontroller adc , sends on serial appreciated.

feel free ask if there's need specify. 

best
andreas

i've seen examples of people doing fft on arduino. signatures may run space problems. sample @ 44.1 khz each sample take 22.6 (1/44100). won't have time except store it, , have 2048 bytes of ram. assuming need few of other things, can therefore store 2000 * 22.6 of samples,, 45 ms of audio. is, under 1/20 of second of it. doubt useful signature out of tiny fraction of incoming audio

Arduino Forum > Using Arduino > Audio > best practice for hi-q ATM168 audio ADC

arduino