Note : The following article contains information that has been deemed incorrect by leading digital audio engineers. I attempted to corroborate the findings of this article by asking several digital audio experts. I was unable to find anyone who could back up the statements made, with any scientific data or theory. Consider the following article retracted.
I am leaving the text of this article up on CA because it has enabled a good discussion to take place. By leaving it up, people can read what was claimed and read the followup arguments that the prove it incorrect. To remove the article completely only opens up a space for this to happen again, and again, and again.
I take full responsibility for the publishing of this article. I should have had a technical editor check it before publication. I apologize to the CA Community for the error in judgement.
Temporal Domain of Signal, or What is More Important for Listening to Music, Static or Dynamic Characteristics of the Sound Signal?
Every time my audiophile friends, who do not have an analog setup (TT), come to me and see huge piles of expensive, rare LPs, they get puzzled. They wonder, how can it be that LP lovers spend huge amounts of money on their "analog" hobby, while suffering such discomforts when listening to music. They say this method of listening in the 21st century is absolutely impractical. In addition, there are signal distortion and limitations in many of the technical aspects of vinyl.
In response, I always say the same thing in support of analog - it's mainly because of the time domain signal. We (fans of analog audio) are willing to make these sacrifices and inconveniences for much better performance in a time aspect, the so-called dynamic characteristics. Static characteristics, those belonging to the spectral and dynamic domains (Dynamic Range, THD + N, Frequency Response, etc.) certainly are important for high-quality sound, but when it comes to listening to music in real time, in my opinion, it is the dynamic characteristics that matter most
Often, in response to my comments, people react with skepticism. They say they are used to trusting technical information that can be measured and compared and what I say is very subjective and ephemeral.
Also viewing comments here on СA, especially those connected with the current topics such as MQA, I have noticed that some members react rather skeptically to the arguments about MQA's improvements of characteristics in the time-domain. And, some even question the very existence of such improvements.
Here it is shown that "High-resolution in temporal, spatial, spectral, and dynamic domains together determine the quality value of perceived music and ,sound and that temporal resolution may be the most important domain perceptually". Temporal resolution, is actually what I would like to briefly discuss with you.
There's a deeply rooted opinion that frequency above 10 kHz, and moreover above 20 kHz, contains a small amount of music information. And yet research shows that, for example transients from cymbals contain significant frequency components extending even above 60 kHz. The trumpet playing fortissimo has transients components up through 40 kHz, and in the case of the violin even temporary frequency of 100 kHz occurs.
As you can see quite a lot of music information is contained in frequencies above 20 kHz. Of course, immediately a question is raised: "Are we able to hear it?". To answer this, it is worth mentioning some rarely discussed issues. Commonly cited audibility up to 20 kHz frequency is derived from conventional hearing tests, which are based on the audibility of simple sounds. But there is an alternative look at the issue from the more "dynamic" side. This is the temporal resolution of the ear, not the "static" harmonic content and audibility of pure sinusoidal tones.
This may be more appropriate in a case of music signals than the prospect of simple tones. The actual music signals have a very complex structure as a result of the imposition of the attack and decay of many instruments. More importantly, their frequency spectrum is very different between the short period of the initial attack, or the rise of sound, eg. as a result of pulling a string or striking a key of a piano, and the subsequent, much longer sound decay.
There is a large group of instruments, which are characterized by a very "transient," dynamic nature of the initial attack phase of the sound. Xylophone, trumpet, cymbals and striking a drum achieve dynamic levels in between 120 and 130 dB within 10 ms or less. One thing we can say for sure, it is not possible for a CD-quality sample scattered at 22.7ms to have an opportunity to correct the commissioning attack phase of musical instruments, which are half the distance between two consecutive samples.
And the attack phase is very important for audio reception. In experiments, in which the samples were of wind instruments dissected in a way that combined the short attack phase of one instrument with a longer sound decay of another one, listeners identified the sound that of the instrument with the short fragment attack, not the longer decay sound.
The sound wave graph from a cymbal being struck by a stick. The sound increase is nearly instantaneous, followed by a long sustain of a rather uniform nature. - from highfidelity.pl
When viewed from the hearing mechanism perspective, you can find information indicating that the signals which have pulsing character (i.e., generally transients), in contrast to simple tones, activate significantly larger areas of hearing cells than pure sinusoidal tones (which, in nature are almost non existent). In the case of pulses, the possible temporal resolution of the human ear may be up to 10 microseconds, corresponding to frequencies of 100 kHz.
This information is also confirmed in the opinion of recognized practitioners. Art Dudley from "Stereophile" magazine, in an interesting interview from The Editors cycle, is of the opinion that the Nyquist frequency does not apply while there are working decimation and reconstruction filters of complex music signals. In his opinion, two samples may be used to describe a single frequency, but do not provide sufficient density samples to describe the speed at which the signal increases or decreases. This is crucial to distinguishing between music and ordinary sound.
Also I would like to quote, in the context of the above information, an excerpt from my correspondence with Dr. Rob Robinson:
"My thoughts are that with extended frequency response you are not capturing "audible" frequencies but rather preserving the critical time relationships in the music at all frequencies. Human hearing might not be able to "detect" sounds above 15 - 20 kHz or so, but on the other hand hearing, in conjunction with the brain, is very sensitive to temporal information. It's been reported that the human auditory system is capable of discerning temporal differences of tens of microseconds or less (and note, at 192 kHz the time between samples is 5 microseconds). This temporal discrimination is the reason we are able to accurately discern directional / spatial cues. Hearing evolved so that the location of threats, e.g., the cougar about to pounce, could be determined accurately, as key to survival. The spatial information comes not only from amplitude, but the time difference between the same sound arriving at each ear. And the more sensitive hearing is to temporal information, the more accurately that spatial cues can be located.
A CD format brickwall filter will affect time relationships, part of the reason that CD format digital audio may sound less "natural" than analog (or live sound). Preserving temporal information is key to preserving lifelike sound and imaging. While all digital audio will affect temporal information, the influence diminishes the higher the sample rate, because the antialiasing and reconstruction filters are operating at ultrasonic frequencies. So, by using higher sample rates, even though we may be recording sounds that are inaudible, we have better preservation of the temporal information in the signal, which conveys a more lifelike presentation of the music. Besides using a high sample rate to capture the signal, we also have the ultra wide 5,000 kHz bandwidth (five thousand kilohertz, as contrasted with "just" 20 kilohertz as the generally accepted audible upper frequency limit) of the Seta preamplifier which again faithfully preserves temporal relationships in the music signal (internally, the front end circuitry has a risetime of less than 50 nanoseconds)." - Dr. Rob Robinson
If we take into consideration the typical technical parameters of audio, which is mainly bandwidth and dynamics (signal-to-noise ratio), we can easily come to the conclusion that, omitting the variables associated with the physiology of hearing, audiophile devices should not differ from each other, and moreover sonically stand out in relation to the audio devices from the mass market.
And yet, there are people willing to pay much higher prices for equipment and the typical specifications are often similar or even slightly worse than the cheaper devices of the mass segment.
Most importantly, in many cases audiophiles agree on the description of the main attributes of the sound of the given device, although expressed in a specific descriptive dictionary, and not in strict technical parameters.
This raises a difficult to challenge conclusion that if some audiophile characteristics are consistently perceived by a large number of people there's a good chance that behind this stands specific physical phenomena, though their nature can be complicated and can be difficult to express in simple numerical parameters, eg. dynamic range or frequency response.
What may these phenomena be? If the key to the mystery lies not in the parameters of the frequency domain (frequency response) or dynamics (noise at a low level), then a single area remains, and that's phase issues, or timing aspects of the sound. In fact, these are the most fundamental parameters of the sound signal, because they underlie its creation, what a sound wave actually looks like in the time domain. The question is how much of the sound wave graph corresponds to the wave reaching the microphone registering this recording.
The nuances of the tonal colors, to the greatest extent, are shaped by the sound wave characteristic from each instrument. And, it's not just a simple analysis of the contents of the so-called harmonics but more of dynamic aspects, mainly the so-called attacks, or the rising of sound at the moment of its creation. It is not difficult to imagine that the course of the rise in amplitude of the sound will be quite different for wind, string and plucked instruments. It's a very fine structure of transients, which over a very short period time, this new tone of a musical instrument provides the bulk of information about its color and texture. Studies show that the human ear is most sensitive to the initial part of the pulse of a new musical sound.
Any disturbance or contamination of this sensitive time structure leads to a noticeable loss of sound quality from the perspective of people sensitive to audiophile aspects, such as nuances in fidelity transmission of all the colors of musical instruments.
In other words, the time domain signal (issues phase, or timing aspects of the sound). In fact, these are the most fundamental parameters of the sound signal, because they lie in its creation - thus what a sound wave in the time domain actually looks like.
So, one of the main advantages of vinyl is the lack of restrictions of temporal resolution in LP. One of the key challenges for us in the Pure Vinyl Club was to find a way (technology, a method of recording) the equipment to maintain a maximum level of temporal resolution from the LP while recording in digital. This does not mean that we were going to compromise or neglect other characteristics which are also important for the sound.
Paweł Piwowarski in his article "PLIKI HI-RES - niezbędny krok do nirwany czy nadmiarowy gadżet?" on High Fidelity.pl in the October 2016 issue to which I referred above, noted that "The trumpet playing fortissimo contains transients of 40 kHz". I invite you to watch this little video using our LP rip, which clearly shows that transients of the trombone can get higher than 50 kHz, and trumpet reaches almost 70kHz!
Later, in one of the following articles, which might be called "What is actually recorded on LP" I will showcase many interesting videos and screenshots, which clearly show that in many musical instruments transients exceed the 40-50 kHz threshold, and among them will be some unexpected ones (contrabass and sibilance of the human voice).
Also, many audiophiles have prejudices about the LPs Dynamic Range. Here's a screenshot of the DR of an album's full side (Duration: 24:07, RAW Record).
I will focus on these and other interesting LP aspects in more detail in in the next articles of the Digital Vinyl series.
Trippin (Kenny Drew – Trippin (1984, Japan) Promo WL, Baystate (RJL-8101))
Official DR Value: DR13, Gain Output Levels (Pure Vinyl) – 14.00dB, Edit “Click Repair” – yes
Play Fiddle Play (Isao Suzuki Quartet + 1 – Blue City (1974, Japan) Three Blind Mice (TBM-24))
Official DR Value: DR13, Gain Output Levels (Pure Vinyl) – 8.02dB, Edit “Click Repair” – yes
Make Someone Happy (Carmen McRae – Live At Sugar Hill San Francisco (1964, USA) Time Records (S/2104))
Official DR Value: DR14, Gain Output Levels (Pure Vinyl) – 7.23dB, Edit “Click Repair” – yes
Early In The Morning (John Henry Barbee, 1963
VA – The Best Of The Blues (Compilation) (RE 1973, West Germany) Storyville (671188))
Official DR Value: DR14, Gain Output Levels (Pure Vinyl) – 10.63dB, Edit “Click Repair” – yes
La Cumparsita (Werner Müller And His Orchestra – Tango! (1967, USA) London Records (SP 44098))
Official DR Value: DR11, Gain Output Levels (Pure Vinyl) – 0.00dB, Edit “Click Repair” – yes
Wild Is The Wind (The Dave Pike Quartet Featuring Bill Evans – Pike’s Peak 1962 (RE 1981, USA) Columbia (PC 37011))
Official DR Value: DR12, Gain Output Levels (Pure Vinyl) – 10.31dB, Edit “Click Repair” – yes
People Are Strange (The Doors – 13 (1970, USA) Elektra (EKS-74079))
Official DR Value: DR11, Gain Output Levels (Pure Vinyl) – 0.00dB, Edit “Click Repair” – yes
Let’s Groove (Earth, Wind and Fire – Raise! (1981, Japan) CBS/Sony (25AP 2210))
Official DR Value: DR15, Gain Output Levels (Pure Vinyl) – 7.89dB, Edit “Click Repair” – yes
Smooth Operator (Sade – Smooth Operator (1984, Single, 45rpm, Japan) Epic (12・3P-581))
Official DR Value: DR13, Gain Output Levels (Pure Vinyl) – 7.15dB, Edit “Click Repair” – yes
Fernando (Paul Mauriat – Feelings (1977, 45rpm, Japan) Philips (45S-14))
Official DR Value: DR14, Gain Output Levels (Pure Vinyl) – 5.76dB, Edit “Click Repair” – yes