• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

AKM new flagship AK4191 + AK4499EX

I don’t think we should spread fud/myths about different chips sounding remarkably different . Implementation is everything if the chips are of recent vintage not some relic of the 80’s or 90’s ?

Well built modern DAC’s exceeded our human capabilities long ago . Possibly exception is if you are young enough for excellent hf hearing then the weirdest filter choices with roll off on the top and ultrasonic leakage migth sound different to you .

Please don’t attribute texture detail or soundstage to a DAC , if such words cary any meaningful information it’s about the recordings
I think what you are saying is accurate, and I do not wish to throw in more BS as it may technically be BS or hold no validity. But what I have read is that even the number of languages you know help you hear different. Not that I am a master, but I know 3 langauges and one of them is I think the #1 or 2 most characters in the alphabet which, according to this article I read, helps you pick up sounds better as you are using more of the audible frequencies. Clearly this is a very limited understanding as the talking band is a very small part of the frequency range. So I can see this as BS. Maybe slightly increased with male and female band, if I understand correctly men and women speak in different bands

However, I will throw in and I am not implying this as I have golden ears or any of that, my current and previous paragraph. But I do not know if you have heard about micro noises, but I remember for me this was a popular prank to pull in classrooms when I was in high school. They would play this micro noise and the teacher would never hear because they are old but everyone else who is clearly young is able to pick it up and it's super obnoxious. To get to my point, I am 35 and I still am able to hear that micro noise which is something that was said should not be possible. I may be wrong about the age this disappears in hearing, but I do recall it's something you lose in your 20's

But I totally agree with you on implementation and it seems more than often any time whether it's an audio show or personal units I have owned I did not like ESS. Yet again as stated, one of my streamers has an ESS dac in it and it actually sounds fairly nice. I notice the same issues but it's basically toned down. As in it's still present, but the amount of it getting to my ears is far less which is fine to me.

Precisely as you say, I have owned a couple BB chips and most the cheap units sounded very warm. I have the Ifi Neo IDSD and that one sounds much closer to AKM and realistically I feel it's even a bit more neutral than AKM. So your comments about implementation are true, but again also the chips do vary and I am not sure how much of an influence that may have. This is also why I stated I am still very interested in owning the Gustard X26 Pro to further understand your point, but I think the streamer I own is establishing that as well.

Next we must consider how many possible ways are there to implement a dac, I would guess 3 or 4 max and that would mean sound/experience should be similar amongst those implementations and a dac from the same manufacturer especially if what you point out that dacs sound hardly differ. There has to be a certain limit or available knowledge of implementations unless someone creates a completely new radical idea. Simply based on when you go to these manufacturer websites you see they give you the possible electronic routes these dacs need or have in diagrams

The only thing I would assume gives high variability but should not be a major impact on the sound is component quality in that line. Op amp rolling being one of these and it maybe similar with other parts of the chain. But even than if we are considering that dacs hardly sound different, how much of difference can op amps make. We do see burson for example with its many op amps and they definitely change the sound, but lets be clear I am not even certain if dacs even have op amps. I believe they do.

The thing which gets me and probably further validates your point and some of what I am saying as well. The ESS dacs which was a 500 dollar stand alone, housing duel 9038's non pro, sounded worse than a streamer which implemented a single 9038Q2M in differential configuration and both cost 500. You would clearly assume that the dac itself in a streamer would have a poor implementation as cost is being divided into other things ie included streamer. So more money or resources into implementation would have been used in the stand alone dac. Who knows maybe the company was just being cheap and trying to rip people off I do not know and that's why it came out poopy

I also always get a test routine when I buy a dac to see how it sounds on speakers and headphones, and similarly when it sounds good with speakers it will be good with headphones. Equally with what doesn't sound pleasant to me, but I still give the unit a chance as it may create something I can put to use somewhere.

Who is to say, this may all be placebo or even maybe something I pick up which someone else may never notice.

Edit: Also wish to point out that I am not challenging your statement, nor do I have the expertise to give facts about the matter. I am only discussing my personal experience which has only been 2 years in audio. So I am in not standing to tell anyone they are right or wrong and my statements are for rebuttal which will help me learn or acknowledgement that some things maybe correct or reasonable.
 
Last edited:
Indeed a nice option at that price point :)
I wonder with there will be a DX7 pro plus velvet on the horizon .

Stacks are nice but I'm in the market for a all in one unit :)
Mentioned this earlier, just as the D90 originally had the 4499 maybe an E90 will drop with the new dacs. Would not doubt it with the way topping drops dacs
 
Next we must consider how many possible ways are there to implement a dac, I would guess 3 or 4 max and that would mean sound/experience should be similar amongst those implementations and a dac from the same manufacturer especially if what you point out that dacs sound hardly differ. There has to be a certain limit or available knowledge of implementations unless someone creates a completely new radical idea. Simply based on when you go to these manufacturer websites you see they give you the possible electronic routes these dacs need or have in diagrams
The only thing I would assume gives high variability but should not be a major impact on the sound is component quality in that line.
I feel you assume a lot of things that 'should' make and explain all these sound differences you believe you hear, if you'd have your BB, ESS and AKM DAC devices calibrated to the same volume level and one would switch to device A, B, C; it would be extremely unlikely you'd still go "Ah yes shrill ESS! warm BB! neutral AKM!' you know. All these sighted tests are a whole lot of make belief. There's literally no basis to them.

The whole of AKMs 'Velvet' sound marketing is basically just capitulating on this (make believe) image of a 'softer' more analogue sound. Whatever marketing dude came up with it was hopefully rewarded because it certainly seems to work on.. well.. some people.
 
Last edited:
I think what you are saying is accurate, and I do not wish to throw in more BS as it may technically be BS or hold no validity. But what I have read is that even the number of languages you know help you hear different. Not that I am a master, but I know 3 langauges and one of them is I think the #1 or 2 most characters in the alphabet which, according to this article I read, helps you pick up sounds better as you are using more of the audible frequencies. Clearly this is a very limited understanding as the talking band is a very small part of the frequency range. So I can see this as BS. Maybe slightly increased with male and female band, if I understand correctly men and women speak in different bands

However, I will throw in and I am not implying this as I have golden ears or any of that, my current and previous paragraph. But I do not know if you have heard about micro noises, but I remember for me this was a popular prank to pull in classrooms when I was in high school. They would play this micro noise and the teacher would never hear because they are old but everyone else who is clearly young is able to pick it up and it's super obnoxious. To get to my point, I am 35 and I still am able to hear that micro noise which is something that was said should not be possible. I may be wrong about the age this disappears in hearing, but I do recall it's something you lose in your 20's

But I totally agree with you on implementation and it seems more than often any time whether it's an audio show or personal units I have owned I did not like ESS. Yet again as stated, one of my streamers has an ESS dac in it and it actually sounds fairly nice. I notice the same issues but it's basically toned down. As in it's still present, but the amount of it getting to my ears is far less which is fine to me.

Precisely as you say, I have owned a couple BB chips and most the cheap units sounded very warm. I have the Ifi Neo IDSD and that one sounds much closer to AKM and realistically I feel it's even a bit more neutral than AKM. So your comments about implementation are true, but again also the chips do vary and I am not sure how much of an influence that may have. This is also why I stated I am still very interested in owning the Gustard X26 Pro to further understand your point, but I think the streamer I own is establishing that as well.

Next we must consider how many possible ways are there to implement a dac, I would guess 3 or 4 max and that would mean sound/experience should be similar amongst those implementations and a dac from the same manufacturer especially if what you point out that dacs sound hardly differ. There has to be a certain limit or available knowledge of implementations unless someone creates a completely new radical idea. Simply based on when you go to these manufacturer websites you see they give you the possible electronic routes these dacs need or have in diagrams

The only thing I would assume gives high variability but should not be a major impact on the sound is component quality in that line. Op amp rolling being one of these and it maybe similar with other parts of the chain. But even than if we are considering that dacs hardly sound different, how much of difference can op amps make. We do see burson for example with its many op amps and they definitely change the sound, but lets be clear I am not even certain if dacs even have op amps. I believe they do.

The thing which gets me and probably further validates your point and some of what I am saying as well. The ESS dacs which was a 500 dollar stand alone, housing duel 9038's non pro, sounded worse than a streamer which implemented a single 9038Q2M in differential configuration and both cost 500. You would clearly assume that the dac itself in a streamer would have a poor implementation as cost is being divided into other things ie included streamer. So more money or resources into implementation would have been used in the stand alone dac. Who knows maybe the company was just being cheap and trying to rip people off I do not know and that's why it came out poopy

I also always get a test routine when I buy a dac to see how it sounds on speakers and headphones, and similarly when it sounds good with speakers it will be good with headphones. Equally with what doesn't sound pleasant to me, but I still give the unit a chance as it may create something I can put to use somewhere.

Who is to say, this may all be placebo or even maybe something I pick up which someone else may never notice.

Edit: Also wish to point out that I am not challenging your statement, nor do I have the expertise to give facts about the matter. I am only discussing my personal experience which has only been 2 years in audio. So I am in not standing to tell anyone they are right or wrong and my statements are for rebuttal which will help me learn or acknowledgement that some things maybe correct or reasonable.
I guess we have to convince @amirm to introduce the MacCali micro noise test in his test repertoire.:rolleyes:
 
Regardless the new chips are probably awesome and we will se very good implementations and some not so good , we see when the test starts to stack up :) .

Marketing yes its needed to sell , and they succeeded both AKM and ESS . For some reason we all know what our DAC's uses and people fuss about it in AVR's ?
 
I feel you assume a lot of things that 'should' make and explain all these sound differences you believe you hear, if you'd have your BB, ESS and AKM DAC devices calibrated to the same volume level and one would switch to device A, B, C; it would be extremely unlikely you'd still go "Ah yes shrill ESS! warm BB! neutral AKM!' you know. All these sighted tests are a whole lot of make belief. There's literally no basis to them.

The whole of AKMs 'Velvet' sound marketing is basically just capitulating on the marketing of this image of a 'softer' more analogue sound. Whatever marketing dude came up with it was hopefully rewarded because it certainly seems to work on.. well.. some people.
I can agree with you on this since I definitely have never done a blind test. But I did clearly state that this may all be placebo, not to cover my butt but rather may be the honest fact.

I bought the D30 pro because of Amir's review and I got some fairly decent time in with that dac on numerous equipment and felt it was still not to my personal liking. But on the opposite side I got the D70s and that's by far my favorite dac or maybe I should say unit. That I actually tried to purchase a second one to incorporate into my second stereo setup.

So far everything may be pointing to that, and yes it's basically an assumption based on what I am going off of but I am trying to make a educated guess on the situation. Even though I am not an engineer who understands dac production, so it may not be educated at all. Hence why I said this is strictly my opinion, and not an absolute.
 
I guess we have to convince @amirm to introduce the MacCali micro noise test in his test repertoire.:rolleyes:
lol I may tout it as a micro noise, but reality is it's probably a sound very high up on the spectrum. Clearly as we get older we lose the ability to hear the upper registers and this is a known fact.

I am definitely pointing out that I am not saying this by any means makes me a golden ear or have any type of advance ability to hear something.

I guess overall it would just be a hard pressed issue to really point out anything when clearly an audio chain does have so many components.

Also recall, the main reason why I said anything in the first place is the member asked or mentioned he just bought the regular E70 and now he may regret not getting the E70V.

We cant say for certain anything definitively, but we do see topping and many other manufacturers having a single unit with different dacs. Maybe it's for more sales based on marketing. But it could also be possible that they too feel that there's something that creates a change. Also a great topic to touch upon, as some may assume the implementation of the E70 and E70V should be fairly similar. We cannot say it for certain, at least I cant, but if that's the case for instance we now know the only difference would be the dac and that would be a great way to blind test and get some results. The results maybe inconclusive, however if there's a distinguishable difference than we can say for certain the dac does have some type of signature. Which is the only thing we are trying to conclude right now

As mentioned, this is purely learning experience for me, and I would definitely be glad to be wrong and also told why I am wrong so I can better understand and not just regurgitate garbage and create confusion for others. But i do not expect it as it's clearly a technical subject
 
To better understand this separate architecture, when listening to DSD, the 4191 chip is not involved and the DSD signal is sent directly to the 4499EX ?
For example with a Gustard A26 ("DSD direct" option) with a DSD stream from HQPlayer.
 
To better understand this separate architecture, when listening to DSD, the 4191 chip is not involved and the DSD signal is sent directly to the 4499EX ?
For example with a Gustard A26 ("DSD direct" option) with a DSD stream from HQPlayer.
It should be a passthrough when in direct mode, yes.
 
It should be a passthrough when in direct mode, yes.
Look here: https://www.akm.com/eu/en/products/audio/audio-dac/ak4191eq/

1675630396671.png


Output of AK4191EQ is 7bit SDM signal MBD1 ... MBD7 (thermometer code). DSD is 1 bit only. So with DSD input AK4191EX does some kind of 1 bit to 7 bit signal conversion. The question is which data rates between the two chips are supported.

AK4499EX diagram is here: https://www.akm.com/eu/en/products/audio/audio-dac/ak4499exeq/

1675631019062.png
 
"Input Support" only means that the signal will be played back, not necessarily that it is passed on without processing.

I think the most direct path you can hope for is through DSD filter 2 (DSDD bit = 1), which presumably has 7-bit output. 7-bit at 256fs is generally more valuable than 1-bit at 1024fs (without noise shaping, it would take oversampling a 1-bit DAC by a factor of 2^7 = 128 to match a 7-bit DAC, or a ludicrous 32768fs in this case), and the filter is likely to eliminate anything past 256fs/2 so you can easily undersample at this stage. Or that's what I would do, anyway. It's not like we're very interested in anything past 16fs even at DSD1024.

It appears the old (pre-fire) AK4499 kept on just scaling up to DSD512, filter and everything. DR and THD took a hit at DSD512 though (131 / -106 dB instead of 134 / -124 dB).
 
without noise shaping, it would take oversampling a 1-bit DAC by a factor of 2^7 = 128 to match a 7-bit DAC
2^7 would be needed if the 7 bit interface between the chips would be binary weighted. But delta signal modulator output is multibit thermometer code.
Then, we need to count with 2ch stereo signal.
The chip could simply pass last 7 bits of incoming DSD stream to output pins MBD7 - MBD1.
7 bits of right channel, then 7 bits of left channel and so on - at 256fs. So each channel multibit data would be effectively output at 128fs.
At DSD1024 input rate 8 bits of both channel data can be accumulated during 1/128fs time period.
 
. But delta signal modulator output is multibit thermometer code.
No. Conversion to thermometer code happens at the DWA stage. The DS modulator output lines are binary coded.
 
There is firmware for the R26 (1.3 >1.4). I suppose there is no official firmware for the A26 but is the firmware only for the streamer and written to the internal SD card (or does it also update the dac management menu).
And if it only updates the streamer, is the R26 firmware A26 compatible ?
 
No. Conversion to thermometer code happens at the DWA stage. The DS modulator output lines are binary coded.
Hm, thanks. I am interested to know how much is DSD signal manipulated (in the DSDD bit "1" case).

On AK4499 (not EX) block diagram nothing appears between delta sigma modulator and switched resistor DAC. Although the block diagram could be simplified. Do you know what's the difference between older AK4499 and the new two chip solution?

1675705156052.png
 
Hm, thanks. I am interested to know how much is DSD signal manipulated (in the DSDD bit "1" case).

On AK4499 (not EX) block diagram nothing appears between delta sigma modulator and switched resistor DAC. Although the block diagram could be simplified. Do you know what's the difference between older AK4499 and the new two chip solution?

View attachment 262884
It's just a simplified block diagram.
Today's DS-DACs all use an array of equal switched cells (capacitors, resistors, current sources, whatever) for silicon process related reasons. So, for a, say, 5-bit DS modulator output you need 2^5 = 32 equal elements. The binary code of the modulator must be transformed into the number of elements to be switched on -- that's what the "thermometer" coding stage does. Since the matching between the elements is not perfect, the DWA (digital weight averaging) is introduced which pseudo-randomly selects which actual cells from the array are used to fulfil the requirement of how many cells need to be switched on at a time. This reduces distortion (trades it for increased broadband noise).

The whole idea of DSD is to avoid a DAC proper, all you need (theoretically) is an output logic gate followed by a low-pass filter. But the requirements for the voltage and timing precision for that logic gate are insane when you want high precision (low noise, low distortion). So, in a DAC chip which has a DSD bypass to the final output DAC stage (the array of cells) one could use simply use two fixed binary codes (of same magnitude but opposite polarity), make them thermometer code and apply DWA and feed the cell array with that. But I'm not sure if that's really what's going on...
 
Back
Top Bottom