Recording is easy. Dump the ADC output to a buffer and call it a WAV file. Or an AU.
Sending it to Siri requires chunking the audio into small pieces, compressing them with a wavelet table (I'm assuming here, my DSP knowledge is a bit dated), and shipping them off to North Carolina.
If this can't be done quickly, Siri appears to suck. Apple is religiously conservative about enabling features on hardware that might cause the experience to suck.
Audio compression was a slow thing on desktop computers when desktop computers had similar specs to the first iPad. I'm no authority, but this seems like a relevant detail to me.
Recording is easy. Dump the ADC output to a buffer and call it a WAV file. Or an AU.
Sending it to Siri requires chunking the audio into small pieces, compressing them with a wavelet table (I'm assuming here, my DSP knowledge is a bit dated), and shipping them off to North Carolina.
If this can't be done quickly, Siri appears to suck. Apple is religiously conservative about enabling features on hardware that might cause the experience to suck.
Audio compression was a slow thing on desktop computers when desktop computers had similar specs to the first iPad. I'm no authority, but this seems like a relevant detail to me.