No, that's just what a steep IIR (minimum phase) reconstruction filter will do. This not only reduces group delay down in the audio band but, given that there can't be any pre-echos related to periodic passband ripple, you can also get away with much more of said passband ripple than I'd otherwise be comfortable with as well, as any post-echos would be swamped by psychoacoustic masking. Group delay would increase substantially when approaching cutoff frequency, but it's not like human hearing cares a whole lot about phase, let alone near 20 kHz.
Beyond a certain minimum performance level, FIR (linear phase) and IIR reconstruction filters should sound the exact same. Human hearing should be more tolerant towards IIR, while measurement applications would generally favor FIR.
I tend to be taking observed audible differences between DACs with a medium-sized salt shaker. Most people don't even manage anything near the requisite <0.1 dB level match (or <0.3 dB at the very least), at which point further discussion becomes pointless.
BTW, just in case anyone should be using a DAC8 stereo as a plain line-level DAC rather than a preamp replacement, better stay away from the top 2-3 dB of volume control range if you are feeding it bit-perfect audio data from typical commercial CDs and the like with no further digital level reduction. (Now why anyone would be treating a DAC with 131 dB(A) worth of dynamic range - more than a good many analog preamps - as an ordinary line-level job is beyond me, but whatever. Pearls before swine and all.)
