Jump to content

My DAC used to support higher audio quality before I installed Realtek drivers

D1Wolfgang
Go to solution Solved by homeap5,

Double post, because this will be a little different answer.

 

Everyone knows (in theory) how it works, but for those who don't - I prepared small image:

816.gif.cfb70cac991ca795462c979e834c89d2.gif

On left image you see 8bit sampling (small fragment of full diagram) - horizontal values from 0 to 4 (full range is from -128 to +127). Vertical lines are points where sample of sound is taken (sampled). So horizontal lines (0, 1, 2, 3, 4) is bitrate, vertical - sample rate. We will focus on bitrate here.

 

As you see in 8 bit sampling analog curve is changed to digital values with very low precission, because there are no values between, for example, 2 and 3.

 

On second image (16 bit) you can see fragment of first image but in 16 bit. As you can see - it's not, like someone may think, 2 times better precission. It's 256 times better precission! In 8 bit resolution you can have only values 2 or 3 (second sound sample), since in the 16 bit - values from  768 to 1024 - 256 times more!

Third image it's only zoom of second image to show that precission.

 

16 bit precission is perfect - you probably never get DAC, aplifier, cables and speakers that can reproduce it that accurate. And imagine that you have 256 times better precission than 16 bit - you'll get 24 bit. Or imagine that between 16-bit value "1023" and "1024" you have 65536 values - you'll get 32 bit precission. Do you really believe that any amplifier, DAC, speaker or even cable can reproduce it that accurate?

 

If 32 bit is so good and accurate, then why different amplifiers and speakers sounds different? 32 bit sound should be so perfect that should play the same as natural sound. But not - it plays different, because of different analog conversions with different precission. If you hear that differences (and for sure you can hear difference between good speakers and bad speakers), then it's for sure not 24 vs 32 bit difference or even 16 vs 24 bit difference, because you simply cannot hear that.

 

http://www.thrash.pl/816.wav

 

Here you can download some example that I've made - wav that changes bitrate 4 times. Please tell me which part (1, 2, 3 or 4) is 16 bit and which one is... 8 bit! Yes, it's 16 vs 8 bit. If you hear no difference between 8 and 16 bit, then you for sure will not hear difference that is 65536 times smaller than this one (24 bits vs 32 bits).

 

PS. Some of my explanations may be not clear because my english may be not good enough to explain technical things.

 

 

 

When I first plugged in my dac, I was able to set the audio to like 32bit, 96000Hz. Now, after installing some Realtek Drivers from my motherboard manufacturer's website, it only goes as far as 24bit.

I'd kinda like the 32bit back, even though I probably couldn't hear a difference. But more so, I want to know why this happened. I'm very curious since I don't see how Realtek drivers could affect something they aren't driving. 

DAC: FX-audio X6

Thoughts become things. What you hold in your mind, you will hold in your hand. 

Be happy for yourself.

My rig :) :

Spoiler

 

PCPartPicker Part List: https://pcpartpicker.com/list/tX4hWD

CPU: AMD - Ryzen 7 2700X 3.7 GHz 8-Core Processor  
Motherboard: Asus - Prime X470-Pro ATX AM4 Motherboard 
Memory: Corsair - Vengeance RGB Pro 16 GB (2 x 8 GB) DDR4-3200 Memory 
Storage: Samsung - 860 Evo 500 GB 2.5" Solid State Drive   
Storage: Seagate - Barracuda 2 TB 3.5" 7200RPM Internal Hard Drive  
GPU: Gigabyte - GeForce GTX 1080 Ti 11 GB AORUS Video Card 
Case: Fractal Design - Meshify C ATX Mid Tower Case  
Power Supply: SeaSonic - FOCUS Plus Gold 650 W 80+ Gold Certified Fully Modular ATX Power Supply  

 

Operating System: Microsoft - Windows 10 Home OEM 64-bit  
Monitor: AOC - G2770PQU 27.0" 1920x1080 144 Hz Monitor 
Keyboard: Corsair - K70 RGB RAPIDFIRE Wired Gaming Keyboard 
Mouse: Corsair - Sabre RGB Wired Optical Mouse  
Custom: Beyerdynamic DT 770 PRO, 250 ohms 

 

 

Link to comment
Share on other sites

Link to post
Share on other sites

The Cirrus Logic CS4398 DAC used on it is 24bit 192kHz. The receiver on this unit is Cirrus Logic CS8416 which is 32bit 192kHz. Which means unit can accept up to 32bit 192kHz audio, but can only output up to 24bit 192kHz. It's fancy to have 32bit processing, but ultimately, I don't think it'll make any difference in quality.

Link to comment
Share on other sites

Link to post
Share on other sites

No difference in quality at all. Also 192kHz gives no difference over 44.1kHz. These numbers are for marketing purposes mostly. And for people who thinks that they hear difference (while in fact they mostly hear distortions produced by it). But good luck with your "golden ears". :)

Link to comment
Share on other sites

Link to post
Share on other sites

Double post, because this will be a little different answer.

 

Everyone knows (in theory) how it works, but for those who don't - I prepared small image:

816.gif.cfb70cac991ca795462c979e834c89d2.gif

On left image you see 8bit sampling (small fragment of full diagram) - horizontal values from 0 to 4 (full range is from -128 to +127). Vertical lines are points where sample of sound is taken (sampled). So horizontal lines (0, 1, 2, 3, 4) is bitrate, vertical - sample rate. We will focus on bitrate here.

 

As you see in 8 bit sampling analog curve is changed to digital values with very low precission, because there are no values between, for example, 2 and 3.

 

On second image (16 bit) you can see fragment of first image but in 16 bit. As you can see - it's not, like someone may think, 2 times better precission. It's 256 times better precission! In 8 bit resolution you can have only values 2 or 3 (second sound sample), since in the 16 bit - values from  768 to 1024 - 256 times more!

Third image it's only zoom of second image to show that precission.

 

16 bit precission is perfect - you probably never get DAC, aplifier, cables and speakers that can reproduce it that accurate. And imagine that you have 256 times better precission than 16 bit - you'll get 24 bit. Or imagine that between 16-bit value "1023" and "1024" you have 65536 values - you'll get 32 bit precission. Do you really believe that any amplifier, DAC, speaker or even cable can reproduce it that accurate?

 

If 32 bit is so good and accurate, then why different amplifiers and speakers sounds different? 32 bit sound should be so perfect that should play the same as natural sound. But not - it plays different, because of different analog conversions with different precission. If you hear that differences (and for sure you can hear difference between good speakers and bad speakers), then it's for sure not 24 vs 32 bit difference or even 16 vs 24 bit difference, because you simply cannot hear that.

 

http://www.thrash.pl/816.wav

 

Here you can download some example that I've made - wav that changes bitrate 4 times. Please tell me which part (1, 2, 3 or 4) is 16 bit and which one is... 8 bit! Yes, it's 16 vs 8 bit. If you hear no difference between 8 and 16 bit, then you for sure will not hear difference that is 65536 times smaller than this one (24 bits vs 32 bits).

 

PS. Some of my explanations may be not clear because my english may be not good enough to explain technical things.

 

 

 

Link to comment
Share on other sites

Link to post
Share on other sites

Honestly those numbers are cool and all but it's generallybreccomended you stay that the bitrate and frequency your music is generally playing at. 24 bit @ 44.1 khz is basically the best we can produce right now. 

Link to comment
Share on other sites

Link to post
Share on other sites

A high end dac like the Benchmark dac 3 only has around 21 bits of resolution. So even in the high end dac market, the 24 bit format isn't a bottleneck.

 

I have no idea how audio drivers for one device could impact what formats are available on a separate device.

Link to comment
Share on other sites

Link to post
Share on other sites

3 hours ago, an actual squirrel said:

A high end dac like the Benchmark dac 3 only has around 21 bits of resolution. So even in the high end dac market, the 24 bit format isn't a bottleneck.

 

I have no idea how audio drivers for one device could impact what formats are available on a separate device.

Pretty sure it's a percieved bottleneck purely on snake oil marketing and higher numbers on DAC means DAC is better right? Kinda thinking.

Link to comment
Share on other sites

Link to post
Share on other sites

2 hours ago, an actual squirrel said:

A high end dac like the Benchmark dac 3 only has around 21 bits of resolution. So even in the high end dac market, the 24 bit format isn't a bottleneck.

 

I have no idea how audio drivers for one device could impact what formats are available on a separate device.

In practice, yeah, it's hard to think of a "normal" music file that would benefit from more than 16 bits of resolution. But purely talking "bits" of resolution, humans can hear signals well below the noise floor, so it's a difficult argument to make based on SNR alone. At normal listening levels in my current environment I can reliably make out a 1kHz tone 36dBrms below white noise (I could probably go further with more practice and a quieter setup but that's enough to make the point). Since well-designed electronics should be dominated by thermal noise, this implies that with enough amplification even a 21 bit ENOB DAC would be able to produce audible tones beyond 24 bits.

 

Not that it matters in practice. Normal people don't listen to music loud enough in an environment quiet enough for this to be audible.

Link to comment
Share on other sites

Link to post
Share on other sites

2 hours ago, Nimrodor said:

In practice, yeah, it's hard to think of a "normal" music file that would benefit from more than 16 bits of resolution. But purely talking "bits" of resolution, humans can hear signals well below the noise floor, so it's a difficult argument to make based on SNR alone. At normal listening levels in my current environment I can reliably make out a 1kHz tone 36dBrms below white noise (I could probably go further with more practice and a quieter setup but that's enough to make the point). Since well-designed electronics should be dominated by thermal noise, this implies that with enough amplification even a 21 bit ENOB DAC would be able to produce audible tones beyond 24 bits.

 

Not that it matters in practice. Normal people don't listen to music loud enough in an environment quiet enough for this to be audible.

You can pick out constant tones that are under a random noise floor, but outside of contrived situations, I don't think this is that applicable to the real world.

Link to comment
Share on other sites

Link to post
Share on other sites

5 hours ago, an actual squirrel said:

You can pick out constant tones that are under a random noise floor, but outside of contrived situations, I don't think this is that applicable to the real world.

Music and voice and pretty easy to make out too. Human hearing works off spectral integration, so as long the noise is uncorrelated it's not difficult to hear a signal.

Link to comment
Share on other sites

Link to post
Share on other sites

24 bit is enough.  I would argue 44.1 kHz is not enough, I would go with 48kHz.

In theory, 44100 Hz is enough, because human ears are unlikely to perceive anything higher than around 19kHz, and you need double the frequency to correctly reproduce audio, so 2 x 19 = 38 which is below 44.1.  Anyway, 48kHz can help with some high frequencies and it's not an extra effort for the sound card or Windows mixer and resampling 44.1 to 48 is not hard.

 

As to the original question, I would guess that the 32bit format was just floating point values instead of plain 24 bit signed sampled, so you didn't get increased resolution or higher quality, just a different way of passing the samples. Padded 32bit samples is also a thing - basically take 24 bits and just add a null byte to it.

I'd suspect the older WHQL or whatever driver you had in Windows was doing the conversion between 32 bit and 24 bit signed integers so applications could see 32 bit and send 32 bit samples to the driver but the driver would then send 24 bit samples to the actual sound chip.

 

Yeah, even the highest end DACs and amplifiers won't really do more than 21-22 bits of precision. Proper 24 bit DACs for audio are expensive, in the tens of dollars, and they also need high grade voltage references to actually be consistent and produce those bits. 

Link to comment
Share on other sites

Link to post
Share on other sites

3 hours ago, mariushm said:
 

Yeah, even the highest end DACs and amplifiers won't really do more than 21-22 bits of precision. Proper 24 bit DACs for audio are expensive, in the tens of dollars, and they also need high grade voltage references to actually be consistent and produce those bits. 

when you said expensive i thought you meant like audiophile expensive and we were going to have an argument. you had me in the first half

Link to comment
Share on other sites

Link to post
Share on other sites

Thx all.

Fyi, I literally saw the option to set it higher than it is now. I saw 32 bits @ 96000k. I even had it set like that for a bit, but it changed randomly, or after I installed other audio drivers. So even if it is snake-oil or just for advertisement, I was curious why the option went away. 

No biggie. All good info, thank you.

Thoughts become things. What you hold in your mind, you will hold in your hand. 

Be happy for yourself.

My rig :) :

Spoiler

 

PCPartPicker Part List: https://pcpartpicker.com/list/tX4hWD

CPU: AMD - Ryzen 7 2700X 3.7 GHz 8-Core Processor  
Motherboard: Asus - Prime X470-Pro ATX AM4 Motherboard 
Memory: Corsair - Vengeance RGB Pro 16 GB (2 x 8 GB) DDR4-3200 Memory 
Storage: Samsung - 860 Evo 500 GB 2.5" Solid State Drive   
Storage: Seagate - Barracuda 2 TB 3.5" 7200RPM Internal Hard Drive  
GPU: Gigabyte - GeForce GTX 1080 Ti 11 GB AORUS Video Card 
Case: Fractal Design - Meshify C ATX Mid Tower Case  
Power Supply: SeaSonic - FOCUS Plus Gold 650 W 80+ Gold Certified Fully Modular ATX Power Supply  

 

Operating System: Microsoft - Windows 10 Home OEM 64-bit  
Monitor: AOC - G2770PQU 27.0" 1920x1080 144 Hz Monitor 
Keyboard: Corsair - K70 RGB RAPIDFIRE Wired Gaming Keyboard 
Mouse: Corsair - Sabre RGB Wired Optical Mouse  
Custom: Beyerdynamic DT 770 PRO, 250 ohms 

 

 

Link to comment
Share on other sites

Link to post
Share on other sites

4 hours ago, D1Wolfgang said:

Thx all.

Fyi, I literally saw the option to set it higher than it is now. I saw 32 bits @ 96000k. I even had it set like that for a bit, but it changed randomly, or after I installed other audio drivers. So even if it is snake-oil or just for advertisement, I was curious why the option went away. 

No biggie. All good info, thank you.

I've only seen few devices capable of 32bit 96kHz. Mostly it was Creative stuff based on Core3D chip and I think some high end C-Media audio chips also support 32bit. Pretty much everything else is 24bit.

Link to comment
Share on other sites

Link to post
Share on other sites

3 hours ago, RejZoR said:

I've only seen few devices capable of 32bit 96kHz. Mostly it was Creative stuff based on Core3D chip and I think some high end C-Media audio chips also support 32bit. Pretty much everything else is 24bit.

It's more common than you think my d10 which is like a $80 unit has the capbilities to go 32 bit 384 khz. Weirdly enough the d30 only does 24 bit . The khadas tone board also does 32 bit 384. The smsl Sanskrit also does 32 bit 96 kHz . And the smsl m100 does 32 bit 768 khz. And these are only sub $100 options

Link to comment
Share on other sites

Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now

×