Can spectrogram lie about audio quality?

[Zerow]

@stevehero

Just to show you.

I've stopped using Spectro on the account of it not being able to utilize my resolution of 1440p- it is just to small on my screen

@stevehero

Compared to Fakin' The Funks redition. Look at the resolution of the higher frequencies and the cutoff. It's what's called a lossy master.

I've seen such data many times and was suspecting what you just described- a lossy mastering


One symptomatically case [and the final proof to my thesis] was that of music from the 1982 movie "Blade Runner", released as late as in 1994. Wanting to upgrade it from some old MP3s I downloaded one set of FLACs- and they were stopping at ~20+ kHz. So I search for another source- there were the same. So I went to my closet, took out the box with CDs, ripped the original disc [EastWest Records America 4509-96574-2] with Audiograbber on maxed out settings and there it was- a blank space above 20 kHz. Here is an example, the song "One More Kiss, Dear":

And just to make sure I checked the 2007 Special Edition of that score, which on disc 3 contains additional music [not the original compositions]. Here is the track "Up And Running", which samples heavily that "One More Kiss, Dear" song:

It is quite masked with all the new sounds with which the layered the sampled original, but for me to spot such a cut off line is no longer a problem- if I see something like that then I suspect that something like a lossy samples are the culprit. And also comparing different tracks from the same album can tell a lot. That is most evident with OSTs, were you have music from different sources


As for Fakin' The Funk. Here are spectrograms made by it for those same file of "One More Kiss, Dear" and "Up And Running"

For sure I would never spot the issues of samples in the "Up And Running" as I did in Spek. And the "One More Kiss, Dear" was reported on the list of files as a positive [green thumb up]- and I checked the settings [aggressive, report above 256, analyze entire]. So it seems that although FTF has easy interface and can be used for mass check-up, the results may be incorrect

[Zerow]

@Zerow

It is quite masked with all the new sounds with which the layered the sampled original, but for me to spot such a cut off line is no longer a problem

What I wanted to write was

It is quite masked with all the new sounds that are layered over the sampled original, but for me to spot such a cut off line is no longer a problem

[I cannot edit this post no more and I have already flagged this for moderation in hopes of correcting my mistake]

[Zerow]

@Zerow

[...]
Spek can show a frequency cut off line - and so can Audacity. But when Audacity does it- then there is really nothing above that line. But with Spek it is different. Some compressed files [made I do not know how] if are made extremely louder [something like 500-1000%] can show in Spek a little of something than looks like a noise above that supposed cut-off point [that something that you can see in Audacity without making them louder]

On a side note but in relation to issue of checking frequencies:


If anyone wants to checking specific frequencies then you can use Audacity. Under the

Effects > Equalization...

the user can quite precisely wipe out frequencies at any range. Unlike in may other software equalizers, which usually stop at 16 Khz with manipulation of sound, Audacity wipes out at levels as high as the specifications of a given file are [e.g. 96 Khz]. And by that manipulation [in other equalizers] I mean adjustments / tempering and by wiping out [in Audacity] I mean going even as far a as completely eradicating any data

So if you want to check your ability to hear high frequencies and know what really you are missing when listening to CD-quality instead of a digital 96 Khz version of the same recording, then you can wipe out everything below 22 Khz and hear for yourself: [SPOILER ALERT] absolutely nothing

[ohrenkino]

@Zerow

96 Khz

I think that the 96 kHz is the sampling rate.
If you want to sample a given signal, you have to have a sampling rate at twice the maximum frequency of that signal that you want to convert into numeric values.
So if you want to sample 20 kHz, you should use a sampler with 40kHz sampling rate.
If you have a stereo signal then you need that sampler for both channels - so you need a 80 kHz sampling clock.
The 96 kHz not necessarily mean that you find an audio signal with 96 kHz in the file, it tells about the technology used when sampling the original signal.
See also

Sampling (signal processing) |

[Zerow]

@ohrenkino

I think that the 96 kHz is the sampling rate
[...]

You are probably right

But I am referring to marketing schemes when the music industry tries to sell the same old music only because it was remastered and is being sold in XXX Khz; and not mentioning, that it sound different [not always better may I add] because it was mixed from scratch or at least somehow filtered


https://www.youtube.com/watch?v=xH8I0LUjrqw - two guys sit in a car with an unknown hardware setup and talk about how great it sounds, with its greatness coming supposedly only out from a file used and not the hardware. And then the video talks also about feet of water, when trying to diminish CD-quality... hilarious, simply hilarious

[seandex]

@Zerow

I'm sorry to bring the more than 4 years old topic.
Like @Zerow I also found that the upsampling result exposes some hidden area of the sound. and it does sound better also noticeable with the right pieces of equipment.
and I believe that the sound should not be treated as the same mathematics in digital, something always dissolved into the computed digit 010101, and then that codes always could be converted to analog without losing the original analog or almost the same quality.

Everywhere on the audiophile forum, this kind of topic never ends, and so far until 2020, it seems nobody knows the truth. or even though he does, many disagree, as my post will sure get disagrees.

I'm sharing what I found, in the test results, I have been upsampling many of old mp3s and then compared to real vinyl rip, I could not find the big differences after upsampling.

Many people from other forums are saying it is added noise, and I agree, yes that is part of true, some files do make noise after upsampling but that's not all cases, I would say 25% from my experiences. most of them got improves in quality without making the noise.

I have been testing hundreds of different converting, mixing, sampling and I still could not find the real answer for this argument, and still getting confused. it's been almost 10 years, maybe some people have master degree in audio, and some are a real genius or have special ears to judge everything which might be far better than my testing period.

Like how many of us would pass the blind test where the certain bitrate nowadays that didn't even reach the middle quality (not mp3)? There was an audiophile contest in South Korea and surprisingly mostly ALL of them failed at the 160Kbps test at round 3.

It seems noise or something garbage that should not be there after upsampling,
but I love that noise, it sounds like a real instrument to me so I would rather have that noise in all of my collections, therefore, I don't believe it is noises,

and the vinyl sounds always good, and those are 48khz. however, audio scientists (?) is saying humans can't hear above 40khz. Maybe that is right, I don't want to disrespect their testing result or science, so here goes philosophy over their science about missing 8khz.

"Though free to think and act, we are held together, like the stars in the firmament, with ties inseparable. these ties cannot be seen, but we can feel them." -Nikola Tesla

[Cat]

Most of what you said here - without any specific point - is incorrect or confused.
One thing where it is clear:
<< "and the vinyl sounds always good, and those are 48khz. however, audio scientists (?) is saying humans can't hear above 40khz. Maybe that is right, I don't want to disrespect their testing result or science, so here goes philosophy over their science about missing 8khz." >>

The 48 kHz is the sampling rate, sampling frequency.
(You can learn about it at the Wikipedia page. LEARN)

Human hearing range maximum frequency is 20 kHz.
However, very few people >= 12~13 years can hear as high as 20. Most people over about 40yo cannot hear higher than 16 kHz.
(Think of that when you read what old farts on audiophile forums are saying about treble and tweeters.)

So there is no "missing 8 kHz" and when you find some apparent anomaly like that, you should know that the problem is your understanding or your mistake.

You have the option of learning, studying, all relevant information is easily available. But you will not get far without good understanding of Mathematics at least high school [Grade 12] level.

[Cat]

No, spectogram cannot "lie" about audio quality.

Spectogram does not show "quality", it shows sound pressure level (SPL) in decibels (dB) of frequency range along time scale [of an audio track/file.

if you look at many, you will find that it is very unusual to see any energy >16 kHz that is at audible level. Darker blue colour indicates dB level that is too low / too quiet for human hearing.

As software, it is either correctly written or it is not.