Techmoan video: Shaving Compact Discs to improve the sound

[coppice]

Then came flac and ape, which are lossless algorithms but still based on 16bit 44.1KSa/s for as far as I'm aware.

I'm not sure about APE, but FLAC can be pretty much any bit depth or sample rate you like.

[fourfathom]

I think that nowadays it will all be done on a computer where all the data is present on the hard disks and tracks are shifted in time and cut in pieces and then mixed together into a final track.

That's how I was doing it 20 years ago.  Multi-take multitrack digital recording, time synch via click-tracks (not clock sync, this was for human use), some recorded live, some recorded in the studio, others sent in from remote locations (not real-time), etc.  I don't know what process requires all the clock synchronization, but perhaps there is one.

[SiliconWizard]

Then came flac and ape, which are lossless algorithms but still based on 16bit 44.1KSa/s for as far as I'm aware.

I'm not sure about APE, but FLAC can be pretty much any bit depth or sample rate you like.

Of course. I have a ton of 24-bit FLAC files. There's a number of retail sources where you can buy them from. (I'm not talking about pirating. Although you can probably find some stuff using that too.)

[pcprogrammer]

@SiliconWizard. Can you hear the difference between 24bit and 16 bit versions of the same music. I have never tried it, so no idea if I would.

[DavidKo]

I was quite surprised with his procedure. I will read both cd in computer and compare the data. I must say that his approach was very nice.

[rsjsouza]

Techmoan's videos are quite interesting, and this one didn't disappoint. Audiophoolery aside, this was revealing of the lengths people go to try and scam others.

[Zero999]

That video appeared in my recommendations. I smirked, thought what a load of bollocks and didn't bother watching it. I wonder if that happened to anyone else here?

[ataradov]

I think the tests here were not really necessary. They were just for content. There is absolutely no way to reproduce any differences and this is known in advance.

The video is just to show off a goofy device from the past.

[SiliconWizard]

@SiliconWizard. Can you hear the difference between 24bit and 16 bit versions of the same music. I have never tried it, so no idea if I would.

Yep. As long as the material is really 24-bit mastered content, and not just 16-bit content requantized. The former is now available from various resellers.
The main point is that due to the extended dynamic range, audio that was properly mastered @24-bit instead of 16-bit uses much less heavy compression. Excessive compression for 16-bit CD mastering is very common. And that's definitely annoying. Now whether you'll hear the difference really depends on the kind of music that is. For some kinds, it doesn't really matter if there is "heavy" compression or not. For others, it definitely does.

[TimFox]

I recently attended a live concert that featured Olivier Messiaen's "Quartet for the End of Time", which has an extremely large dynamic range (far greater than young peoples' boomedy-boom) between the loudest and quietest sections.
During the lowest loudness sections, the hall's ventilation was clearly audible.  Nevertheless, the details of the acoustic instruments were clearly audible to the unaided ear.
I didn't have my portable sound level meter with me, and it was probably out of range anyway.
With non-sinusoidal musical waveforms, what is the quantitative relationship between the required bit depth and the overall dynamic range?

[DavidKo]

I think the tests here were not really necessary. They were just for content. There is absolutely no way to reproduce any differences and this is known in advance.

The video is just to show off a goofy device from the past.

You are looking on it with wrong optics. You must look at it as you do not know anything about it and want to buy this CD destroyer second hand. Everything written about device itself can be true, except that it is nonsense from the CD point of view. From the optical point of view the black maker on the side will work the same way independently on the angle, even without cutting it. Marker will simply absorb the light in the same amount with very small influence of the angle. On the other hand the wave guide effect itself cannot work, first there is no total reflection on the aluminized part and additionally the light is scattered with the data, so there will be almost no light on the edge, except reading really close to the edge.

By the way most of the Techmoan's videos show devices from the past and this is nice video about scamming people, which happens all the time. I expect that in few year someone will recreate this device and will sell it in the same way (probably including 5 star reviews in magazines).

[AndyBeez]

I agree grinding the edge improves balance, but do compact discs read from the inside outwards? I don't know as I'm no connoisseur de audio.

Anyway guys I need to know this; will angle grinding improve the video signal (SN ratio) from a PWM encoded 1980s LaserDisc? The LaserDisc is a quasi analog system therefore, angling the edge should give near 4K performance from a VHS quality disc. Surley this is just exponential extrapolation ?

[golden_labels]

As for the light absorption, thre is no particular requirement regarding the marker and dye it uses.

Keep in mind CD readers use 780nm light, which is near infrared. Dyes are designed to cover visible spectrum and neither anyone cares if they absorb IR nor you can test that with your naked eye. While one can argue that some pens use blue dye to obtain black color, and that it absorbs red and likely NIR, that can’t be said about pens using red dyes.

So that part can also be considered untrue, even if covering the edge in some paint would provide any benefits.

[NiHaoMike]

Anyway guys I need to know this; will angle grinding improve the video signal (SN ratio) from a PWM encoded 1980s LaserDisc? The LaserDisc is a quasi analog system therefore, angling the edge should give near 4K performance from a VHS quality disc. Surley this is just exponential extrapolation ?

No need to modify the disc, digitize it and run it through Nvidia DLSS. There are a number of examples of just how good upscaling can be nowadays.

[AndyBeez]

No need to modify the disc, digitize it and run it through Nvidia DLSS. There are a number of examples of just how good upscaling can be nowadays.

I also realised old LaserDiscs are too large to fit on this disc shaver  My theory about shaving is, if it really did work, it was because the first generation of CD players did not have the best lasers?

I recently upscaled some old family VHS-C content digitised to MP4. Took about a week of processing time, but the grain from the original camera's CCD is now really noticable sharp. GIGO.

[james_s]

Yes, CDs play from the inside outwards, and no, there was nothing wrong with the lasers in early CD players, the first players worked very well, later ones were mostly just less expensive to build.

[NiHaoMike]

I recently upscaled some old family VHS-C content digitised to MP4. Took about a week of processing time, but the grain from the original camera's CCD is now really noticable sharp. GIGO.

At least in theory, deep learning (that's what DLSS is - Deep Learning Super Scaling) can recognize what's noise and what's detail.

[onsenwombat]

I agree grinding the edge improves balance, but do compact discs read from the inside outwards? I don't know as I'm no connoisseur de audio.

Anyway guys I need to know this; will angle grinding improve the video signal (SN ratio) from a PWM encoded 1980s LaserDisc? The LaserDisc is a quasi analog system therefore, angling the edge should give near 4K performance from a VHS quality disc. Surley this is just exponential extrapolation ?

Remember to switch your laserdisc player power cables to >99.9% silver ones, otherwise your k's roll over and the system might be damaged.

On-topic, what a comedy gold. Been a while since I've had my dose of high-end audiophile lunacy, and this certainly made me chuckle few times.

[coppice]

@SiliconWizard. Can you hear the difference between 24bit and 16 bit versions of the same music. I have never tried it, so no idea if I would.

24 bit is extremely useful in a recording and production environment, as it means you can still get a good 16 bits after adjusting gains, applying various forms of processing, like compression, and so on. However, it is not necessary for distribution. If you can hear the difference between 16 and 24 bits from a distributed recording the 16 bit one must have been poorly prepared. Either the gains were messed up, or they didn't apply dithering properly.

Most recordings, even classical ones, are now much more compressed than in the past. This can be a good or a bad thing. In most situations capturing the full dynamics of an orchestral concert experience in the home means you have to turn up the quiet passages, and turn down the loud ones to avoid disturbing other people. However, even if you apply no compression at all, 16 bits properly dithered will capture all the dynamics your ears can detect.

[coppice]

I recently upscaled some old family VHS-C content digitised to MP4. Took about a week of processing time, but the grain from the original camera's CCD is now really noticable sharp. GIGO.

At least in theory, deep learning (that's what DLSS is - Deep Learning Super Scaling) can recognize what's noise and what's detail.

DLSS is basically "making shit up when there is no genuine information to work with". Just as an artist can take a messy picture and air brush it into something nice looking, so can DLSS. That's good for making things look more pleasant, but there is no actual detail in there. Don't expect anything in the details to be genuine. It can often look "clear" when its totally bogus.

[NiHaoMike]

24 bit is extremely useful in a recording and production environment, as it means you can still get a good 16 bits after adjusting gains, applying various forms of processing, like compression, and so on. However, it is not necessary for distribution. If you can hear the difference between 16 and 24 bits from a distributed recording the 16 bit one must have been poorly prepared. Either the gains were messed up, or they didn't apply dithering properly.

Most recordings, even classical ones, are now much more compressed than in the past. This can be a good or a bad thing. In most situations capturing the full dynamics of an orchestral concert experience in the home means you have to turn up the quiet passages, and turn down the loud ones to avoid disturbing other people. However, even if you apply no compression at all, 16 bits properly dithered will capture all the dynamics your ears can detect.

It's also useful for playback if you want to do something other than just play the file bit perfect. For example, if you apply equalization, some frequencies will have to lose resolution. With 24 bit playback, you'll have a lot of room to keep the usable resolution to 16 bits or more.

[BrianHG]

If you can hear the difference between 16 and 24 bits from a distributed recording the 16 bit one must have been poorly prepared. Either the gains were messed up, or they didn't apply dithering properly.

And that's the kicker.
16bit done right is superb under 99.99% of the use cases.  (Unless you are trying to encode the full authentic dynamic range of gunfire and explosions, however, who has a sound system which can effectively destroy your internal organs...)
I can only, if I tried, make a case for a higher sample rate than 44.1Khz, and it doesn't need to be much higher.

[Fixpoint]

I wonder about that interpolation bit if that also applies for a CD reader in your computer when it reads "digital data" and not "audio data". I assume that it will only do this interpolation when it has read part of the audio data stream, detects that there are errors, tries to resolve the errors, can't and will then interpolate between previous and next samples.

This doesn't make any sense from any viewpoint. It is important to understand what interpolation actually is. In order to make my point simpler, let us assume a very simple form of interpolation by taking the average.

• The "average" is a statistical measure, namely the arithmetic mean, defined to be the sum of data points, divided by the size of the set. However, a non-audio CD doesn't contain any data points, so the average is not defined. It just doesn't make sense to talk about the "average" when working with a non-audio CD because you are talking about things that don't exist/are not defined.
• Maybe you think because any data can be encoded as numbers, this gives us a meaningful definition of an average. That is not correct. The average would only be defined on the encodings but not on the represented data itself. Illustration: When we have a bit sequence representing an "A" and another sequence representing a "B", what does the average of their numerical encodings mean in the data domain? Nothing.

These points extend to all other forms of interpolation.

[pcprogrammer]

First of all there is no mention of average, but interpolation between possibly multiple samples to reconstruct what it originally might have been. There are all sorts of algorithms to do such an interpolation. Does not have to be linear.

Furthermore it was about reading the audio tracks on a computer. The question was about if the computer makes a distinction when reading it as digital data. In other words if a different method is used when playing the CD in your computer or reading it as digital data. In both cases it concerns the audio data, and not some binary text or program.

Edit: you reworded your post, while I was writing my reply

[Fixpoint]

Edit: you reworded your post, while I was writing my reply

Yes, I noticed that I had written my response with the simplest example in mind which could invite people to say "it doesn't apply to the general case". Oh dear. Well ... it does!

There are all sorts of algorithms to do such an interpolation. Does not have to be linear.

Well, what I tried to explain is that this exact statement has nothing to do with the problem at hand, that's the whole point. Linear or not, when talking about interpolation we are talking about statistical methods. For those to apply, you need to data points which we don't have on a non-audio CD.

Furthermore it was about reading the audio tracks on a computer. The question was about if the computer makes a distinction when reading it as digital data.

When the audio CD data is treated as non-audio, you already have given the answer yourself. In this case, the hardware cannot apply any kind of interpolation since there are no data points to interpolate. If non-audio data cannot be read, this is called an I/O error. Hardware that says "oh my, I can't read those bits! but never mind, I'll just make some up real quick" would be useless.

However, AFAIK the hardware can be configured to ignore read errors and continue reading.