DTS:X APO4 + DTS Interactive for Most Devices [USB Supported]

[itry2079]

When the sampling rate exceeds twice the target frequency, increasing the sampling rate does not seem to increase the "resolution".

 

[Ferather]

It increases the interval, which means you get more samples per second. If all frequencies above 20kHz are removed, but the interval stays the same, you get more samples of 20kHz per second.



You can only have 1 total voltage down 1 wire of copper, the speaker can also only be in 1 physical position at a time.

----

Essentially microphone to speaker, fully DMAS (with PCM code limits for amplifying), only the mic can clip, the mic will always have a fixed max volts.
Lets say the mic was also RGB optical (PAM), with 100 positions, and the speaker 100 positions, you get 1:1, regardless of voltage.

The mic could be 5v / 100, whereas the speaker could be 14v / 100, same 100 positions, different volume.
Maximum voltage for a long period will produce a flat line, at max volume.

====

The best way to do modern audio is to forget about analogue and work in bits, you could argue that a speaker is 1 bit time.

 

[itry2079]

I think your theory applies to bit depth, not sampling rate.

Here's his video, I just generated english subtitles for it (.srt file).

Video about sample rate

MediaFire is a simple to use free service that lets you put all your photos, documents, music, and video in a single place so you can access them anywhere and share them everywhere.

www.mediafire.com

In this video, this guy did an experiment:
Two sine waves with the same frequency, one from a 48000hz file and the other from a 96000hz file. One of them was inverted and then played at the same time. The result was no output, indicating that the two waveforms were exactly the same.

 

[Ferather]

Same in that video even with a simple sinewave, if we include more dynamics its more obvious, see PCM in my next image.





If that guy did the same the peaks circled would be seen as lines.

You can also see PCM is a binary form of PAM.

 

[itry2079]

No, the audio processing software connects the sampling points directly with straight lines, which can save CPU resources.
The actual stored waveform is not like this.
Immediately after the time you took the screenshot, there was an explanation in the video.
(The .srt file in the folder is the English subtitle file)

 

[Ferather]

You also need to factor in demodulation accuracy. The higher the total available positions, and positions per second the higher the accuracy (most notable when dynamic).
Imagine A4 graph paper, with 2cm x 2cm squares, and trying to draw an accurate wave from line height in a row, vs 1mm x 1mm squares.

Position (height), sample interval (width). 100 positions (height) 100 positions per second (width).

====

With a certain amount of positions and positions per second, there might not be any need for demodulation, which would be very interesting.



As mentioned previously, you can imagine speakers as bit and bit time, position (voltage) at position time (interval) even demodulated.

 

[Arctucas]

Only two place, two "{Device-ID}"

Yes, there are two places where "Device-ID" appears.

I am asking; do I delete all the listed device ID and replace them with the copied GUID from my card?

Or, do I add my GUID to the existing list?

 

[Ferather]

Delete and replace both Device-ID, with the copied GUID from your card. Examples:

HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Windows\CurrentVersion\MMDevices\Audio\Render\{3d4ee8a3-01b6-49a7-9f50-08dc32425858}\FxProperties
HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Windows\CurrentVersion\MMDevices\Audio\Render\{3d4ee8a3-01b6-49a7-9f50-08dc32425858}\Properties

Where {3d4ee8a3-01b6-49a7-9f50-08dc32425858} is my device GUID.

 

[itry2079]

Both 48000hz and 96000hz sampled waves can theoretically restore the original waveform. Two sampling points within half a cycle can restore a sine wave (Shannon's theorem). Different waveforms can be decomposed into sine waves (Fourier transform).

Of course, in reality it is related to the D/A conversion of the sound card.
So I think demodulation is mainly related to hardware parameters (sound card, amplifier).
I guess the high sample rate might make up for the shortcomings of the lower end hardware.

 

[Ferather]

Most audio is not a simple sinewave, I admit if the demodulation is expecting a sinewave sure it can guess the line between two intervals, and possibly be no different with more intervals.
If there is a change between the intervals, the demodulation will be inaccurate as there are no intervals to represent it, so guessing its a sine would be wrong.

To be honest nobody plays a simple sinewave at 1 frequency, music-other is multiple frequencies at once.




The higher rate can represent the curve between.

====

Two DAC's with the same input interval, and output voltage, both sound different, none are lossless. Might have 0.0001% THD (sinewave), why?
Its also possible to be accurate with 1 frequency sinewave, then inaccurate with another (THD), why?

As a side note Class-D and PWM, uses 4096 x input frequency or up to 200Mhz (not Khz) for samples, so 48k x 4096.

----

AM (amplitude modulation) and FM (frequency modulation), (both analogue) and a demodulator are worth a Google.
You will soon notice its easier to work with digital and bit, bit time opposed to analogue specs.

====

Pulse Code Modulation and Demodulation : Block Diagram & Its Working (elprocus.com) (PCM = PAM).
Binary is limited to 1, and 0, it takes a lot of bits to represent 1 position, unlike RGB Optical.

 

[itry2079]

Limitations of PCM​

The sampling theorem like Nyquist–Shannon illustrates the operating of pulse code modulation devices can be done without establishing distortions in their frequency bands if these bands offer a sampling frequency as a minimum twice that of the maximum frequency included within the i/p signal.

I'm a little confused.
So will the sampling rate setting in Windows adjust the quantization resolution of the hardware device?
In my mind, this setting is software related.

=== === === ===

Since at the same sampling rate, waves with higher frequencies have fewer sampling points.
Can it be said that at a 48000hz sampling rate, the distortion degree of the 17000hz wave is always much higher than that of the 500hz wave?
Then the high-frequency parts of a song should sound fuzzier than the low-frequency parts.

But in fact a sine wave around 17000hz doesn’t sound like a triangle wave.

 

[Ferather]

Yes audio has a resolution, as much as pixels in a graph. Position mapping is grid-graph. Else why bother with 4096 x 48k in PWM Class-D?
Another way to say it, 48k sample rate is not enough resolution, rate, to skip the demodulator (normally a DAC).

PAM: 25 channels, 4608k sample rate: 25 (C) x 4608000 (S) = 115,200,000 / 1,000,000 = 115.2 Mbits/s

Traditional PCM (100 bit) would need: 11.52 Gbits/s to do the above, no RGB.

----

A copper wire can only be 1 volt spec at a measured time, the speaker can only have 1 physical position at a measured time.
You can consider speakers as digital, but high rate. Bit, and bit time (position, positions per second).



PCM - PWM (not PAM), no DAC.

----

The only filter in a DMAS system (PAM) is unwanted, but captured frequency from the microphone (over 20kHz).



Very direct (basic example), you use the off state for no power.

====

Note I have limited my calculations based on 125 Mbits/s TOSLink. This value can be much higher.
For example a RGB transmitter and receiver might do 20 Gbits/s as maximum.

Number of channels and rate are limited by the number of colours-lumen, colours-lumen p/s.

 

[itry2079]

You gave a clear answer. It seems to require quite a lot of bandwidth.
My sound card uses a USB2.0 interface, which theoretically has 480Mbps, but the actual rate is 120Mbps~240Mbps.

=== ===

Since at the same sampling rate, waves with higher frequencies have fewer sampling points.
Can it be said that at a 48000hz sampling rate, the distortion degree of the 17000hz wave is always much higher than that of the 500hz wave?
Then the high-frequency parts of a song should sound fuzzier than the low-frequency parts.

But in fact a sine wave around 17000hz doesn’t sound like a triangle wave.

How to explain this phenomenon?
It stands to reason that at a 48000hz sampling rate, a 500hz sine wave will be much smoother than a 17000hz one (500hz wave has more sampling points in a half cycle).
The 17000hz wave has only about 3 sampling points in a half cycles, but it still "sounds smooth".

(forgive my bad drawing)

 

[Ferather]

Depends how well demodulation is going (DAC's all differ). The red lines between intervals has no actual information to go on other than previous point (interval) and next.
The red line its self shows us how fast an interval is needed to run without a demodulator, speakers respond to the generated red line (voltage).

Without information between interval, its guess work based on pervious and next position.

====

Welcome to the optical age, non-binary. Infinite bit specification means that 100 bit and 1000 bits are equal in transmission (same bit rate).

PAM: 25 channels, 4608k sample rate: 25 (C) x 4608000 (S) = 115,200,000 / 1,000,000 = 115.2 Mbits/s

Could be 100 bits (100 positions), or 1000, will still be 115.2 Mbits/s.

----

With enough positions (bits) and intervals (positions per second), we can go direct microphone to speaker, no amp, without loss or change.
Even if there was some demodulation, the accuracy would be so high, any 'guess work' would be undetectable.

Considerably more direct (extremely low latency also), with considerably less parts.

What is True Sound? The concept explained | Stuff

----

The more pixels on your screen, the more accurate a line and curve can be drawn, without looking like squares (modular).

====

Here we see a basic example of my 'open' (not copyrighted) DMAS unit, and speaker. The PCM audio processing portion (RGB-DSP) will be working with [255,255,255,X] code (RGB-Lumen).

You could consider the PCM portion as advanced PCM, as the 'binary' equivalent of 'RGB-Lumen' code could be reduced in size as a [default format].
Similar to the way float processing works opposed to fixed. The amp can be coded to never go higher than [X] position.



DMAS = Digitally Managed Audio System (RGB Optical, PAM X).

 

[itry2079]

Thanks bro, you are really professional.

With enough positions (bits) and intervals (positions per second), we can go direct microphone to speaker, no amp, without loss or change.
Even if there was some demodulation, the accuracy would be so high, any 'guess work' would be undetectable.

That will require very high hardware specifications and transmission bandwidth.

DMAS = Digitally Managed Audio System (RGB Optical, PAM X).

LOL, I work in the embedded field, and I always thought DMAS = Direct Memory Access System.

 

[Ferather]

Not much bandwidth is needed with RGB. As mentioned sinewave is easier to do than dynamic audio. RGB allows us to do much higher accuracy at a minimal cost.

Funny how much information shows on Google with some cookies.

Binary is a limiting factor, digital is not.

 

[itry2079]

I discovered an interesting phenomenon.

Keep "Volume Smoothing" (similar to loudness equalization) off.
"Treble Enhance" or "Dialog Clarity" as you like, no matter how you set it.

The focus is on "Bass Boost":
1. Keep its switch on;
2. First set the slider to a higher value (for me just > 20);
3. Then **quickly** slide it to the 0 position (You can also click directly on the left side of the progress bar, as long as it is the 0 position. It must be 0 and cannot be greater than 0!);
4. You can clearly feel that the sound field has become larger. The larger the value in step 2, the more obvious the effect.

---- ----
I tried using EQ to boost the high frequencies, but no matter how I adjusted the EQ, I couldn't achieve this effect.
It sounds more spacious and more realistic now, and I'm curious how this is done.
Putting aside the principles of programming for a moment, I wonder whether the sound we hear in real life is very different from the recorded sound.

Sounds of different frequencies attenuate to different degrees when propagating, which can cause some problems.
For example, when a movie voice actor is recording a slice, he is very close to the microphone. But if in real life we keep a certain distance from the person speaking, the sound will attenuate.
That is to say, the recorded audio and the real sound have different proportions of high, medium and low frequencies.
Sound recorded in a studio, while clearer, may be less realistic. (There is also the reason for space reverberation, we have mixer, ignore it for now)

---- ----
In comparison, high frequency seems to account for more in real life.

Low-frequency sounds decay more slowly during propagation, and high-frequency sounds decay faster, which seems to contradict the above.
I second thought about it, not really. In reality, the sounds we hear are often those that have been reflected. High-frequency sound waves are more reflective.

 

[Ferather]

Interesting, I never use analogue myself I just test all features work. Noted, that may change in the update.