I'd like to comment on a couple of posts to this list. First, on 1/23 Corey Baily wrote:
" I would like to re-word Ted's comment and say: 'Not all analog EQ exhibits the same phase characteristics'.
This is the basis for the timeless argument of why some prefer the sound of one brand of analog preamp (phono or otherwise) over another.
Those that have suggested that digital EQ is much better at modeling the phase characteristics of analog preamps are correct. However, the
software engineer has to first pick which analog phase characteristics that he or she wants to use for the algorithm.
So, the argument of which sounds the best continues!"
OK. This is a complicated issue, and background that follows is a bit tedious. But, it's necessary to illustrate my point.
Nearly all phono EQ curves are a combination of 2 or 3 first-order (i.e. 6 dB/octave) filters (there are just a couple of odd-ball exceptions). Two in the case of most older curves that did not define a low-bass shelf, and 3 for the revised NAB curve, the Columbia LP, and RIAA. While there are multiple phase characteristics for the myriad higher-order analog filters (Butterworth, Bessel, Chebyshev, to name a few), there is only ONE amplitude and phase characteristic for a first-order filter, because a first-order filter contains one resistive and one reactive component, and there's only one way the components can be configured for either a low-pass or a high-pass characteristic. All records cut with an RIAA curve have the same phase characteristic, and all phono preamps providing accurate RIAA playback equalization have THE SAME phase characteristic, and that phase characteristic is a mirror of the recording characteristic. This equally true for any other recording and complimentary playback characteristic, whether 300 Hz / -5 dB, AES, NAB, Columbia LP, or any others).
I lived through all of the debates that began back in the 1970s about whether or not all phono preamps sound the same. Back then, and for at least two decades after that, the prevailing view among the engineers who were running the AES was that the only audible differences were frequency response, level and absolute polarity (I call these the MAAD parameters, for Mothers of All Audible Differences). If those characteristics were the same then all amplifiers, all preamps, indeed all audio electronics, would sound the same to the human ear. By the late 1970s, one thing the "golden ears" and the "scientists" generally agreed on was that minor differences in frequency response were, indeed, audible. This led to the view that phono playback equalization needed to be super-accurate, in order to ensure that frequency response could be ruled out as a source of audible differences between them. It was, and I think generally still is, agreed that if an RIAA phono preamp is accurate from 20 Hz to 20 kHz, +/- 0.1 dB, frequency response can probably be ruled out as a source of audible differences. I agree with this. It's important to remember that, although the conventional wisdom is that the ear cannot distinguish level differences of less than 1 dB, the nature of phono equalization errors is that they are normally spread over two octaves or more, and listening to program material (as opposed to test tones), the ear is extremely sensitive to a depressed or elevated frequency response if its spread across a region of an octave or more. This is especially true if those errors are across the midrange portion of the spectrum. If an RIAA error elevates the region between 2 kHz and 8 kHz by even a few tenths of a dB, that error will be readily audible on program material. Hence the agreed-upon need for extremely accurate phono equalization, whether RIAA or any of the older curves.
Now, I know there are those who will say that the component tolerances in the filter networks the produced those older curves probably had tolerances of 10 % on a good day. This is, no doubt, true. But, what do you want to hear, what's actually on the record or errors in your phono preamp?
Now, to get to the punch line. One of the main points in my 2010 article is that, with analog filters, phase response is a direct result of the frequency response produced by the filters. The phase response "comes along for the ride." Two identical analog phono equalization filters, with exactly the same frequency response, will also have exactly the same phase response. Phase response differences between analog phono equalization networks are a byproduct of frequency response differences between those networks. Phase response differences are not the cause of audible differences in phono preamps, beyond the differences in frequency response that produced them in the first place. To be perfectly clear, it is not possible for two phono preamps to have identical frequency response, yet different phase response.
So, the statement "'Not all analog EQ exhibits the same phase characteristics" is incorrect insofar as it is applied to phono equalization networks designed with first-order filter characteristics. If a software engineer designs an algorithm to mimic the phase response of analog phono equalization networks, there are no "choices" when it comes to phase response. For first-order filters, there's only one. The phase curves I show in my 2010 paper are the only ones possible for the turnover and roll-off points I'm illustrating. There are no other choices.
OK, if I haven't driven everyone to drink with that explanation, allow me to move on to Tim's email:
"I understood that phase shift as a result of normal EQ in a mono track, or the same phase shift applied to both channels is inaudible. But phase shift can be audible when applied to only one channel of a stereo track, as it affects the stereo image, because we perceive a source's location partly through the arrival times of the signal to each of our ears."
What I think Tim is saying here is that if phase shift is identical in two stereo channels playing back a monaural recording, that phase shift will be inaudible. But, if phase shifts are different in the two channels, it is audible because it affects the stereo image. Right on the second point - phase differences between the two stereo channels most certainly will shift the image. I discussed this in the conference paper I have last spring on digital phase correctors and their effects on stereophonic recordings. Stereo recordings are dependent upon differences in both amplitude an phase. Change the phase relationship between the two channels and you will certainly change the stereo perspective.
So, it's important that the phase relationship of the two channels be identical on a monaural recording in order for the image to be located squarely between the two loudspeakers.
But, what about the audibility of phase shift if it's identical in the two channels? In order to tackle this issue, we must understand the difference between phase shift and phase distortion. This issue was addressed in two AES papers that I cited at the end of my 2010 article. In his 1986 AES paper the late Deane Jensen, founder of Jensen Transformers, showed that if phase changes linearly with frequency, it will plot a straight diagonal line on a graph and WILL NOT cause waveform distortion. If phase shift is non-linear, then it is called phase distortion, and will cause waveform distortion. In their 1982 AES paper, Lipshitz, Pocock and Vanderkooy conducted a series of double-blind tests which showed that phase distortion was audible both on program material and test tones.
The nature of phone equalization curves, whether the recording or the playback curves, is that the phase shifts are non-linear with frequency and, therefore, constitute phase distortion. Look at the phase response graphs in my 2010 paper - they are definitely not straight, diagonal lines! Only some the earliest electrical recording characteristics, which may have had a bass turnover but no treble pre-emphasis, would yield a linear phase response. But, even then it would be over the relatively narrow frequency response of the recording apparatus. Once a non-linear recording curve is corrected with the complementary non-linear playback curve, phase distortion has been eliminated.
The bottom line is that, even though the phase shifts on a mono recording, played back in stereo, are identical in the two channels, if those phase shifts are non-linear, they constitute phase distortion and are definitely audible. On any recording where there was a bass turnover and treble pre-emphasis, the phase response will be non-linear. If the non-linear phase response is left uncorrected, it will be audible.
I realize that part of this conversation has to do with EQ applied for reasons other than mirroring the recording characteristic. What I've addressed pertains to actual phono equalization, and not EQ applied in playback for other, often perfectly valid, reasons.
Audio Engineer Emeritus
The Crane School of Music
SUNY at Potsdam, NY 13676
"Great art presupposes the alert mind of the educated listener."
"A true artist doesn't want to be admired, he wants to be believed."
From: Association for Recorded Sound Discussion List [mailto:[log in to unmask]] On Behalf Of Tim Gillett
Sent: Wednesday, January 24, 2018 6:10 AM
To: [log in to unmask]
Subject: Re: [ARSCLIST] Multiple conversions
I understood that phase shift as a result of normal EQ in a mono track, or the same phase shift applied to both channels is inaudible. But phase shift can be audible when applied to only one channel of a stereo track, as it affects the stereo image, because we perceive a source's location partly through the arrival times of the signal to each of our ears.
----- Original Message -----
From: "Lou Judson" <[log in to unmask]>
To: <[log in to unmask]>
Sent: Wednesday, January 24, 2018 3:06 AM
Subject: Re: [ARSCLIST] Multiple conversions
Adding to my previous comment - with good convertors it is fine, and for
your artistic use no problem!
Lou Judson - Intuiitive Audio
It all comes down to "Adjust All Controls For Best Sound" in the end.
On Jan 23, 2018, at 5:55 AM, Steve Smolian <[log in to unmask]> wrote:
> This is not to adjust recording eq with itzs reciprocal playback eq.
> This would be to add subjective eq, I'm working with many orchestral 78s
> from the early electrical era. Mike placement varies so much from session
> to session that it takes a long time to adjust each set. Knowing that I
> get back and make adjustments later, if needed, is comforting. I'm used
> using parametrics in the analog domain to get the sound of massed strings
> "right" as that also clarifies the distinctive sounds of the solo wind
> -----Original Message-----
> From: Association for Recorded Sound Discussion List
> [mailto:[log in to unmask]] On Behalf Of Ted Kendall
> Sent: Tuesday, January 23, 2018 6:28 AM
> To: [log in to unmask]
> Subject: Re: [ARSCLIST] Multiple conversions
> Not all digital EQ exhibits the same phase characteristics as the
> analogue network, and this is important when applyng RIAA playback EQ, for
> On 23/01/2018 11:09, Lou Judson wrote:
>> I don't know in the ARSC community, but in generalaudio engineering,
> digital EQ can be more accurate and efficient than analog, and can be done
> in faster than real time. Why would you want to do that, I'm curious.
>> Mastering engineers do some work in analog domain, but not just simple
>> Lou Judson
>> Intuitive Audio
>> On Jan 22, 2018, at 9:07 PM, Steve Smolian <[log in to unmask]> wrote:
>>> Using 96/24 throughout, what is present thinking about converting from a
> digital file to analog, equalizing, and reconverting, using 2 computers
> 2 converters?
>>> Steven Smolian
> This email has been checked for viruses by Avast antivirus software.