Neil seems to think (and I agree) that preserving content >24kHz for its own sake is pointless, as it's typically inaudible*. If that's true it's true whether the content actually contains orderly-looking sound in spectral view, or whether it's just noise. So that objection pertains to high SR PCM or SACD. (*Leaving aside the possible audible intermodulation distortion arising from playback of ultrasonic content -- whihc is a good argument for FILTERING IT OUT)
DTS 96/24 ostensibly exists to preserve and restore full FR content (frequencies between 24Hz-48kHz) to DTS encodings of 24-bit, 96 kHz SR audio input. The alternative 'core' DTS encoder takes 48kHz input audio as its top rate (i.e., top frequency is 24 kHz) and accepts 24 bit input. Both have maximum encode bitrates of ~1.5 Mbps. Both are lossy...they are discarding some data. And from what I glean, both are lossy at <24kHz in the same way, because the 'core' DTS is already competent at a 48kHz SR.
So the main advantage I see here is that for a DTS 96/24 encode you can start with an audio file that was recorded at 96kHz SR, making it 'more better' than a 48kHz recording, which itself is 'better' than 44.1 kHz ...but only in terms of relaxed ADC/DAC filter requirements.
For a 96kHz input file, the effective output of DTS 96/24, 1.5Mbps encoding will be audio with bandwith up to 24kHz, plus 24 more kHz of inaudible (unless harmful) bandwidth. The output from a plain DTS 1.5 Mbps encoding of the same input file will be bandwidth up to 24kHz, minus ultrasonic content, as there will have been a downsampling step to convert 96 kHz to 48 kHz.
For a 48khz input file, DTS 96/24 encoding requires upsampling to 96 kHz (padding, no information added), while plain DTS encoding requires no resampling of the input file. No differences otherwise from the above.
The lossy perceptual encoding/decoding of the <24kHz content in all cases will be the same (as I understand things).
Sounds like specsmanship. Am I missing anything?
