I think what you're hearing with 96k is the 24-bit word length. I am not convinced that the
super-high sampling rates capture anything audible above what 44.1 or 48k capture, but I do think
that the Nyquist filtering and other factors make the audible top end sound better. However, many
DACs up-sample 44.1k before filtering and converting anyway. For instance, the Benchmark design, of
which there are many variants, up-samples everything to three hundred and something kiloHertz,
re-clocking so as to strip out jitter, then converts to analog.
Here's a "white paper" about Benchmark's DAC1 approach:
For the DAC2 series, the describe the "improved" system this way:
UltraLock2™ Jitter Attenuation System
UltraLock2™ is an improved version of the UltraLock™ system used in the DAC1 and ADC1 product
families. DSP processing is 32-bits, DSP headroom is 3.5 dB, sample rate is 211 kHz, and
jitter-induced distortion and noise is at least 140 dB below the level of the music - well below the
threshold of hearing. Benchmark's UltraLock2™ system eliminates all audible jitter artifacts.
Up-sampling and over-sampling DAC designs have been around for a long time, but I do think modern
designs are more sophisticated in how they strip out jitter from the source. The consumer high-end
designers first got the jitter-rejection religion, especially when they started recognizing consumer
demand for USB interfaces (USB is notorious for jitter due to inconsistent clocking built into
typical computer CPUs). Companies like Benchmark and Mytek and Lynx, which have feet in both
consumer and pro audio, have put out well-reviewed and good-sounding, to my ears, jitter-rejecting
products in recent times. The other focus where I think some strides have been made recently is the
analog stage after conversion, there are some super-quiet and near-transparent designs out there
now. A modern digital system should operate so quietly that it essentially has no audible noise
floor in even a quiet real-world room.
A simple test would be to convert some well-known analog material at 96/16 and 48/16 and see if you
hear a difference. Then 96/24 and 48/24, and then compare the 24-bits to the 16-bits. I think that's
where you'll hear the differences.
To my ears, 24-bit makes a difference, especially with "air and space" in something like an
orchestral recording. Just transferring in 24-bit makes a difference, if you've got a good
dither-down conversion system to get to a CD master.
-- Tom Fine
----- Original Message -----
From: "John Haley" <[log in to unmask]>
To: <[log in to unmask]>
Sent: Sunday, April 05, 2015 2:44 AM
Subject: Re: [ARSCLIST] SACD "surprise"
> CORRECTION. When I said "catching a whole octave above 48 kHz in
> frequency," I meant "catching a whole octave in frequency above what is
> captured by a 48 kHz sampling rate." Sorry about that.
> On Sun, Apr 5, 2015 at 2:38 AM, John Haley <[log in to unmask]> wrote:
>> Thanks for posting the NY Times Boulez article, Tom, which could have been
>> entitled "A bunch of famous musicians sitting around kissing up to Pierre
>> Boulez." They remark how "influential" (i.e, famous) he is. That he is.
>> Does that make him a great conductor? Nope. I loved the Gunther Schiller
>> quote. Obviously, Boulez has occasionally succeeded with a piece of
>> music. Like they say, even a stopped clock is right twice a day. And many
>> great orchestras could occasionally deliver a great performance even while
>> ignoring a monkey on the podium.
>> If DGG digital recordings had max resolution of 48 kHz, as you know that
>> is not an appreciable difference from 44.1 kHz. The difference in
>> frequencies (pitches) those sampling rates will capture is the difference
>> between 22,500 and 24,000 Hz. Way up there, that is a difference of only a
>> note or two (think extended piano keyboard). I have never been able to
>> hear the slightest difference between a recording at 44.1 kHz and one at 48
>> kHz. Recording at 96 kHz is a whole 'nother thing, catching a whole octave
>> above 48 kHz in frequency, but also seemingly able to capture more detail
>> based on double the number of samples. Or maybe I should say capture the
>> detail with greater accuracy.
>> Since we routinely make hi-def dubs (at least 96/24) from analog master
>> tapes these days that can sound really great, I have to wonder if, all else
>> being equal, those results will outshine an original digital recording made
>> at only 48 kHz.
>> I am another one who has never felt that your average DGG orchestral
>> recording captured a lot of the sheer excitement of the sound of a great
>> symphony orchestra.
>> On Sat, Apr 4, 2015 at 8:21 PM, Tom Fine <[log in to unmask]>
>>> Hi Mark:
>>> So from what you're saying, I gather that the maximum resolution of that
>>> Boulez/CSO master would be 48/24?
>>> -- Tom Fine
>>> ----- Original Message ----- From: "Mark Donahue" <[log in to unmask]
>>> To: <[log in to unmask]>
>>> Sent: Saturday, April 04, 2015 6:13 PM
>>> Subject: Re: [ARSCLIST] SACD "surprise"
>>> On Sat, Apr 4, 2015 at 10:31 AM, Tom Fine <[log in to unmask]>
>>>> I can't recall if it was Yamaha or Studer digital consoles, but I think
>>>>> you are correct in your descriptions of "4D". being a true DDD system in
>>>>> that the last time anything was analog was when the mic plugged into the
>>>>> console and the mic preamp went to a ADC.
>>>> The DG 4D system was comprised of a stagebox containing custom remote mic
>>>> preamps and Yamaha converters that connected digitally at 24
>>>> to an RTW bit splitter that allowed them to record 24 bit 16 track on a
>>>> Sony3324. The signal was also distributed to the input of a pair of
>>>> DMC-1000 digital consoles. The normal orchestral kit that I would see
>>>> in the states was a pair or three stage boxes with a pair of machines for
>>>> 32 track recording. It was basically modular and could be scaled for the
>>>> All the best,