80s - Zero Digital - Pure Analog - How would that sound if it were recorded today?

UK instrumental band SKY playing electric treatment of Bach's "Toccata and Fugue in D Minor". Group specialised in fusing a variety of musical styles including light rock, progressive rock, classical and jazz. Audio version appeared on the band's second album "Sky 2" released in 1980. The latter was released as a single (under the name of "Toccata") and reached number 6 in the national pop charts, giving the band the opportunity of performing on Top Of The Pops.

Ram Jam - Black Betty

 

The sample rate/frequency has absolutely nothing to do with how 'smooth' the curves are. And regardless, this is digital audio, your sine is never going to be a 'true sine'. No matter what bit rate you use, its still going to look like a graph plotted function when you look at it.

 

Please stop misleading, I can't understand what you mean what the function means and you threw out the number of points on the graphs like?
right?

http://rain.linuxoid.in/fileupload/sound/sine_8372_192-96-44.png

 

This is only one of many graphical representations. You can also represent the same sine tone as a staircase... or as a smooth sine tone curve. I can show you screenshots of the same sine tone at the same frequency in 44.1k that are graphically smoother than the sine tone representation on the 192k track in your screenshot. That doesn't tell you anything about the sound. It doesn't matter at all how it looks ;)

We need to differentiate what we are talking about when we talk about SR. Does a source with higher SR contain more information? Sure! But if - assuming there is a perfect SRC - we record a 1kHz sine tone at 44.1khz and convert it to 96khz, and record a 1khz sine tone at 192khz and convert it to 96khz as well, then these two tones should null if we switch the polarity on one, with the exception of some noise near Nyquist.

For the listener, 44.1k or better 48k (because of the flatter filter) is perfectly adequate. This means that a higher SR brings no advantage for the listener.

The advantage of a higher SR comes into play during recording and mixing.
Either because you need short recording latencies, or because the recorded source material is to be strongly manipulated in pitch or length.
Or because you want to use waveshaping, but for that you only need a high working SR, not a high source SR.

 

Last edited: Sep 6, 2021

96 Khz much much better ¡

 

lol

record a video and tell me that the earth is flat is good stop smoke mari your sp1200 is outdated

 

Yes, these two videos prove my point perfectly, thank you.

 

In fact, Audacity is being misleading, here is what it sounds like when listening through a DAC

Keep in mind you can't hear >20khz & <100dB in practice. So with that in mind,

• aliasing >20k is OK, since you can't hear it, and
• aliasing <100dB is OK, since you can't hear it

thus, FinalCD is (likely) fine in practice.

Conclusion: stop looking at graphs and START LISTENING

 

Last edited: Sep 6, 2021

Which makes me think of another topic like, the compound? Waveform vs Reality..

When you play a note on a real instrument,
it's not emitting perfectly clean Sinusoidal waves..

It will be something more like:



You know what I mean?

I guess in practice all this is factored in, on the Compound Wave that a regular DAW will represent..
This is just another way of using Fourier Transform to represent the information, right..?

But all this harmonic/timbric microdetail exists, giving sound its character/richness,
while adding lots of Complexity to the captured signals..


Now, I never quite understood how all of this is factored in on a digital audio signal,
with seemingly a resulting Compound wave made of 16/24/32bit samples 44.100/X times a second,
which is effectively able to cleanly represent the sum of all its parts with no apparent detail loss..

It's a baffling thing to imagine, and so Fourier Transform can seem a bit like magic sometimes.

But I'd say that could be another factor why ever higher sample rates/bit depths could be necessary,
it's one thing to capture one instrument..

But again capturing an entire Orchestra, I can imagine the Complexity of the signals increases exponentially,
and that may be one reason why having a higher ceiling of quality depth, and time delta precision, could be useful/necessary..

44.100 captures per second is a LOT indeed,
but again reality doesn't know about time granularity, physics occur with infinite smoothness.. (afaik)

And while 24bit precision will be more than enough to capture/represent one/a few instruments,
I can only imagine having a bigger, more detailed signal, could be beneficial on some scenarios,
considering the full timbrical/harmonic richness, and the Complexity that the sum of all the partial waves could require..


(Alas tho, our sound recording/reproduction technology is still a bit rudimentary,
we're using 2 speakers to vibrate/represent full mixes with tenths of instruments..

And we're using microphones with small surface area to capture those sounds,
usually on just one or two single spots, sometimes more like a Piano or a Decca Tree etc..

But you know what I mean, this is just a Compromise between our technical/practical capabilities,
and the fact that with minimal source capture, but with proper adjustment/production of the sound,
we're effectively able to fool our brains, and take the result as a good/real enough representation of sound/music that works for us.. but as always reality goes much beyond.)

Fourier Transform:
https://en.wikipedia.org/wiki/Fourier_transform

Harmonic Analysis:
https://en.wikipedia.org/wiki/Harmonic_analysis

 

Last edited: Sep 6, 2021

If you look back, since when we have been using these techniques and where we are today, we have come very far. You were the Happy people when the radio was invented, and then the phone, you speak in and someone else, far away, answers. The magic of the A / D and D / A converters, you speak into a microphone and in the computer there are zeros and ones and you can send it all over the world via a power line.

After water and food, electricity comes for me ...! Thank you very much, dear protons, neutrons and electrons, you are doing a great job and without you life would only be half as good.

Oh yes, well that we are never satisfied with what we have, otherwise there would be no progress.
Could it be that greed or longing drives us.

 

"Superposition of sinusoidal wave basis functions (bottom) to form a sawtooth wave (top)"

https://en.wikipedia.org/wiki/Convergence_of_Fourier_series

"Vibration of a single normal mode of a circular disc with a pinned boundary condition along the entire outer edge"

https://en.wikipedia.org/wiki/Normal_mode

Also interesting:
https://en.wikipedia.org/wiki/Spectral_density_estimation

 

Examples of 80's purely digital records (DDD - Digital tape recorder used during initial recording, mixing/editing and for mastering):

 

1993, those were the things I couldn't afford, saw them in keyboard music magazines. I copied some data:

Digital Tape Recording System (DTRS) was introduced by Tascam in 1993. It is a recording system for recording studios that allows eight audio tracks to be recorded digitally on Hi8 cassettes.

As with analog video recorders or DAT, the recording takes place in helical tracking, in which the tape is threaded a bit out of the cassette and wrapped around an inclined, very rapidly rotating head drum on which the recording heads are located. Depending on the DTRS model, sampling frequencies of 44.1 kHz, 48 kHz, 88.2 kHz, 96 kHz and 192 kHz can be selected. However, the higher the sampling frequency used, the fewer tracks are available for recording (only two at 192 kHz).

 

Samplerate convertors comparison:

After some additional listening tests, so far sonically the best is iZotope RX8 resampler.

 

Last edited: Sep 6, 2021

maybe he said "you never need this" but i wanted to be exceptionally friedly this tme.

 

are you about to question 60 years of sampling theorem?

a sinewave of 100 Hz, created in or sampled at 192 Khz, has 1920 samples, so that on a 2k monitor you have to zoom in to see the steps.

 

even programmers often run into math traps when calucalating time or length values because they tend to think of samples as the timeline between point A and point B - because this is how it if often presented to us - but in fact, they are only the points.

it also means that it is much easier to properly record a guitar or a car crash (aka "16 bit 32 khz is enough!") than a sinewave.

cosines and white noise are really sensitive material.

 

Well, ofc a sinewave will never be perfectly smooth, just like Pi having no end; it's a mathematical abstraction..

But I'd say the frequencies emmited by a guitar are just as sensitive as a sine/cosine or white noise,
or perhaps more, given the fact that they are emmited irl, by a real physical instrument..

Maybe a sinewave having some small imperfection, or limited resolution (up to a point) wouldn't be that crucial,
because being synthetic/an abstraction our brains can fill-in the gap..

But all imperfections/lack of resolution on a guitar recording will mean Real information being lost..
(and again, think about the limited/rudimentary nature of our recording equipment/methods,
one or two microphones placed on a given spot will never be able to truly capture the thing as it resonates irl)


I don't know the technical intricacies of digital audio at this level,
but it's also possible that given the timbrical complexity, and broad-range semi-chaotic nature of real acoustic signals,
the digital audio capture could in some situations reach, or get closer to the limit of its capacity..

I mean, you "only" have 24bit to spread along, and represent very complex signals with many partial harmonics and sub-harmonics..
How much complexity can you add before those 24bit are not enough to properly represent it?

A 192KHz recording is sampled 192.000 times per second,
and at 24 bit it would have 32.212.254.000.000 sampling points/second (192.000 x 16.777.216)

So theoretically speaking, a 24-bit 192KHz recording will have over 111.455 times the resolution of a 16-bit 44.1KHz recording..

https://en.wikipedia.org/wiki/Audio_bit_depth
https://en.wikipedia.org/wiki/Single-precision_floating-point_format

 

- With the Red Book standard, in addition to the sampling rate of 44.1 kHz, there is also a word length of 16 bits, which enables only 65,530 volume levels over the entire dynamic range of 96 dB. The human hearing, however, has a much finer resolution. Psychoacoustic research suggests resolutions in the range of over 1 million volume levels, which would correspond to a comparable digital resolution of 20-22 bits.

- The not infrequently better sound of the music material on high-resolution sound carriers, such as DVD-Audio and SACD compared to the music material published on CD on average, is not least due to the fact that the responsible sound engineers and mastering engineers were given more freedom for publications on SACD and DVD-Audio, focus on high sound quality than is usual for CD releases.

On the other hand, of course, is the increasing storage requirement and processing effort of files with higher sample rates. It is also sometimes argued that high sample rates from 176.4Khz and up are even detrimental to the sound quality of the playback, since they load downstream equipment such as amplifiers and loudspeakers with unnecessary high-frequency signals and thus degrade their performance.

For me personally it follows that I always try to get a lossless 24-bit version of every recording. I don't care about the sample rate up to 96kHz. If several versions are available, e.g. 48kHz and 96kHz, I prefer a bit of “headroom” and choose the 96kHz version. In addition, due to the file size, I prefer “not too high” sample rates, i.e. 88.2kHz / 24Bit or 96kHz / 24Bit instead of 176.4kHz, 192kHz or 384kHz - not to mention 768kHz. So if there is the same file when downloading in 24/96 or 24/192, I prefer the 24/96 version.

Source/German: https://digital-audio-systems.com/sinn-und-unsinn-von-hohen-sampleraten/