i'm having the exact same problem. i'm using a standard 8-ohm speaker direct to pins SPK- and SPK+. i have added decoupling caps everywhere as recommended, but doesn't fix problem. my chip is on a breadboard, so i don't really have Vss-a (analog ground) and Vss-d (digital ground) isolated, they are connected of course, i tried to get one on one side, one on the other, but it doesn't seem to affect anything.
i suspect the issue is related to these instructions in the Design Guide:
"separate each power line and each ground path individually from device to system. meanwhile, ISD1700 has isolated Vccp, Vssp1 & Vssp2 pads for PWM speaker driver. to minimize noise from speaker driver, it is vital to route an independent trace from each related pin to the supply and ground terminals directly."
but not sure how to implement this on a small breadboard.
**broken link removed**
some other things i've noticed which might give someone an idea -
- when (battery powered) ipod plugged into ANA-IN, feed-through sound is fairly clear and loud, but when recorded/played sounds terrible, not only very noisy but also a lot quieter
- when computer sound-out plugged into ANA-IN, both feed-through sound and playback are terrible (ground loop?)
- when ANA-IN left unplugged, feed-through sound is very noisy
- when recording from iPod with (clean, loud) feed-through, result is audible, albeit crappy. when recording from iPod with feed-through OFF, result is almost inaudibly low in volume.
- on the demo board schematic, the ANA-IN is pulled down to ground with a 5k resistor before the coupling capacitor and sound input line. when i put this in, the noise on the feed-through is silenced but so is the audio - nothing plays through or is recorded.
i also studied the demo board schematic to give me clues on good layout practice, but again not sure exactly how to translate best into breadboard:
**broken link removed**