I am re-posting these comments I recently made on the high res circle just to make sure AVA fans see the info, as follows:
Problems with the AES report on audibility of high bit rate digital.
In the September, 2007 newsletter of the Audio Engineering Society, Brad Miller and David Moran presented a long, detailed, and (what they thought) definitive report regarding the audibility of high bit rate digital audio as compared with standard Redbook. Their conclusion, given their test methods, was that the ability to reliably pick out high bit rate material (in this case 96/24) from standard 44/16 was random. Thus there was no musical advantage to what all of us interested in this are trying to accomplish.
For a long time I was satisfied that they were correct and that their study was foolproof.
However a closer look at their test setup recently reveals that a fatal assumption was made.
Their setup used a "highly regarded" SACD player as the source playing either 96/24 SACD or DVD Audio discs. The analog output of the player was then sent (via an ABX box) to the audio preamp and power amp and speakers. The analog output was also sent to an ADA converter (analog to digital to analog) which converted the analog audio from the SACD player back to 16 bit 44K digital and then back to analog again. This output was also sent to the ABX box.
The user could select the playback from either direct 96/24 from the SACD player, or reworked 44/16 from the ADA converter. Of course which one was being selected was not reveled to the listeners. Thus the true double blind nature of the test.
Meyer/Moran pointed out that if the ADA box was doing any musical damage, that should have tipped the findings even further in favor of the direct analog output from the 96/44 source playing in the SACD player. It did not.
The conclusion then reached was that there was no audible difference between 96/24 and 44/16 digital.
The bad assumption made was that the source SACD player was essentially perfect. Unfortunately, it likely is not.
Since that study was made, we have learned how to intercept the digital data stream from an SACD player ahead of its internal D to A converters, digital filters, and analog output circuits (something in general not allowed by the format).
When we do this, substituting our own DAC for that built into the SACD player, we can realize obviously better musicality. For sure the SACD player's analog audio output is not the living end!
Thus, all this study likely shows is that the analog output of the SACD player used in this test likely was not of adequate resolution and linearity to distinguish the difference between standard and high bit rate digital source material.
To draw any meaningful conclusions, the test process should be repeated using a foolproof high bit rate source and a foolproof way of making the comparisons. Unfortunately, this highly regarded test does not do that.
Best regards,
Frank Van Alstine
PS I have a .pdf copy of the complete test if anyone is interested.
