What's your go to Desktop & portable DACS?

[I miss my iPod]

Hi all,
First post on the forum, I used to listen to music on my iPods to and from school & collage back in the day. At the time I wasn't aware that MP3 was actually lower quality than CD because it just wasn't something I considered back then! At the time I wouldn't have dreamed of investing more than about £50 in some in ear headphones, but, having gotten into Hi-Fi audio over the last couple of years, I realise that I was only scratching the surface. I've tried out Qobuz/Tidal etc & have a new found love for audio quality again!

At the moment I'm using an iFi Audio xDSD Gryphon with a set of Meze 99 Classics and it sounds really good! Seems to have enough features to suit my needs for now. Next headphones I would like to get my hands on are the Meze 109 Pros as I've listened to these and they blew me away, especially for the price.

Are there any other 'Audiophiles' on here? if so what kind of set up are you rockin'? How have you guys found Apple Music vs other streaming platforms that offer higher quality audio file types?

 

[MajorFubar]

To take your last question first...the absolute biggest difference there will ever be, is the quality of the masters, which is something us as consumers cannot impact. The quality of the masters makes an even bigger difference than MP3 vs lossless.

Similarly, I've never heard differences in so-called 'high res' audio that also couldn't be explained away by the same reason. Case in point, I have a version of Rumours on DVD-A that sonically-obliterates my CD. But if I downsample it myself to 16/44 (ie CD quality) it still obliterates the CD, simply because the master it was made from sounds better than the master they made the CD from. Other people have different opinions and are absolutely sold on the marketing hype that they need 24/96 or above. Which is fine, and is not worth arguing over. But I'll just say that one of the funniest things I ever read a few years ago was someone on the AudioShark forum saying how he'd just bought a 24/192 version of Brothers In Arms (EDIT: actually I think was an SACD) and how the higher resolution made it sound so much better than the regular CD. But if it really did sound better, it cannot be because of the resolution, because BIA was recorded on a Sony 16/48 PCM recorder.

Tbh your Gryphon is a really nice piece of kit. Should be absolutely fine driving 109 Pros with its 1 Watt output. There comes a point when, to notice a really tangible difference, you have to move away from an all-in-one like the Gryphon and spend a load on a separate DAC and amplifier. And that's one heck of a rabbit hole because there's so many different permutations depending on how much you want to spend and how portable your setup needs to be. Many headfi enthusiasts like to mate quality DACs with good valve/tube amplifiers.

 

[smirking]

I wouldn't call myself an audiophile, but I do like my sound and I have a first edition EarStudio ES100. It's tiny, ultra-portable, wireless, and terrific.

It only cost me $90 years ago and I was quite stunned at how good it was for so little money. I still don't see much reason to move on from it, but it's getting long in the tooth and I'm really curious about my options in the $200 range. I don't feel the need to chase the ultimate setup... just looking for very good and good enough.

 

[Avatar74]

To take your last question first...the absolute biggest difference there will ever be, is the quality of the masters, which is something us as consumers cannot impact. The quality of the masters makes an even bigger difference than MP3 vs lossless.

Similarly, I've never heard differences in so-called 'high res' audio that also couldn't be explained away by the same reason. Case in point, I have a version of Rumours on DVD-A that sonically-obliterates my CD. But if I downsample it myself to 16/44 (ie CD quality) it still obliterates the CD, simply because the master it was made from sounds better than the master they made the CD from.

This is exactly it. Some years ago I did an analysis across many tracks, using a spectrum analyzer and Leq(A) meter, and found consistently that regardless of format, the tracks that looked like they were mixed and mastered properly, spatially and spectrally (in frequency and amplitude) balanced well, sounded consistently and noticeably good regardless of format.

There was a great primer on digital audio by Monty Montgomery of xiph.org that explains, largely, why there should be no perceptible difference across a multitude of formats in which the signal is band-limited via lowpass filter to exclude any frequencies above the Nyquist limit (the peak of human hearing range which is quite a bit higher than you can consciously perceive).

I use 24/192 DACs for entirely different reasons... because I am producing content in which both analog and digital signal processing chains exist. Consequently, the noise floor is substantially lower such that multiple loops of the signal chain from digital to analog outboard processors and back again cause no materially perceptible loss, and it's a lot easier for manufacturers to design and build filters that don't require steep cutoffs (a point Monty makes in one of his more technical versions of the above video).

I ONLY use a 24/192 DAC in the studio. What's more is that I have studio monitors with a very flat frequency response. You won't find this in consumer loudspeakers and you definitely will not find it in headphones or earbuds... there is NO headphone in existence with as flat a response as even a middling pair of studio monitors. I have this hardware for the same reason NASA tests its fuel tanks to three times the maximum stress they're rated to handle... that ensures that I can prepare a mix with no artifacts or false color introduced at any stage, so that no matter what equipment you, the listener, are using, the output sounds exactly as it should.

I don't use a 24/192 DAC in any other listening context... completely pointless.

 

[MajorFubar]

I don't use a 24/192 DAC in any other listening context... completely pointless.

Annoyingly though, I do find I sometimes I have to succumb to this nonsense in order to get access to the best-quality masters. Same goes for vinyl records. They're technically an inferior carrier to CDs and lossless digital streaming, but sometimes I have to buy an album on vinyl just to get access to a master that hasn't been brickwalled for the earbud brigade. (Just as well I have a higher-end record player.)

 

[MajorFubar]

I use 24/192 DACs for entirely different reasons... because I am producing content in which both analog and digital signal processing chains exist. Consequently, the noise floor is substantially lower such that multiple loops of the signal chain from digital to analog outboard processors and back again cause no materially perceptible loss, and it's a lot easier for manufacturers to design and build filters that don't require steep cutoffs (a point Monty makes in one of his more technical versions of the above video).

Yeah same here. Sounds to me like we're absolutely on the same page. From the recording and mixing perspective it makes absolute sense to use the highest resolution available to you, even if just from the perspective of the great advice your maths teacher gave you to only round your equations down to two decimal places at the very end, to minimise the risk of introducing material rounding-errors further up. It's basically the same reasoning, and is very valid.

But, as proven by Nyquist/Shannon nearly a century ago, 16/44 is the best you'll ever need in the end product, for human ears.

 

[Avatar74]

Annoyingly though, I do find I sometimes I have to succumb to this nonsense in order to get access to the best-quality masters. Same goes for vinyl records. They're technically an inferior carrier to CDs and lossless digital streaming, but sometimes I have to buy an album on vinyl just to get access to a master that hasn't been brickwalled for the earbud brigade. (Just as well I have a higher-end record player.)

Oh god... Don't get me started on brickwall limiters... Even FG-X's dynamic limiter drives me up the wall.

I generally find that as long as an old vinyl exists, there's usually a CD pressing from the early 80s from that same master. Great examples include Off The Wall (MJ), Give Me The Night (George Benson) and some Quincy Jones compilations... anything engineered by Bruce Swedien is god-tier.

Off The Wall has had several versions, including two versions of the CD made from the original and the 2001 remaster. The Phonorecord right (symbol: ℗) date indicates the date that the work was fixed in a particular compilation (meaning here a collection of songs, e.g. an album) and that usually gives you a clue as to whether it's a remaster or the original.

I tend to use iTunes/Apple Music for simplicity so everything gets switched to 16/44 or 24/48 anyway when you're not specifically using a 24/192 DAC (Music will automatically serve you the 24/48 ALAC or 24/96 ADM BWF if no 24/192 DAC is detected and Hi Res Lossless is selected).

Anything I buy in another format I just transcode to 24/48 ALAC or 256 KBps AAC. Even the latter is determined to be acoustically transparent according to AES.

 

[Avatar74]

Yeah same here. Sounds to me like we're absolutely on the same page. From the recording and mixing perspective it makes absolute sense to use the highest resolution available to you, even if just from the perspective of the great advice your maths teacher gave you to only round your equations down to two decimal places at the very end, to minimise the risk of introducing material rounding-errors further up. It's basically the same reasoning, and is very valid.

But, as proven by Nyquist/Shannon nearly a century ago, 16/44 is the best you'll ever need in the end product, for human ears.

Exactly. I used to work for Bell Labs' sister company, Lucent. Even so, I am astonished every day by just how old Shannon-Nyquist Sampling Theorem (1928) is and how simple, elegant and foolproof it remains.

 

[M.Brane]

Still rockin' my iPod Classic 160GB when I can't do Bluetooth. Not looking forward to the day it finally dies.

At home it's TOSLINK into a Benchmark DAC>Bryston 4B>Yamaha NS1000Ms. Don't do a lot of headphone listening, but pretty happy with my Beyer DT880s.

All local files no streaming. I work in an area with very limited cell service.

 

[Ben J.]

What's your go to Desktop & portable DACS?​

Hi. Retired small studio owner here. I still have a nice little setup with nice studio gear in my living room. Since I don't record much anymore, I moved from AD/DAs to DACs, and I was shocked at the audio quality of these small, inexpensive devices often known as "headphone" DACs. Compared to regular audio interfaces that most amateur/semipro studios use, they can often be much better. I've tried a Olasonic and a couple of Topping DACs and they were all great. For now, I've settled with a Topping E30 II. I don't care if it's cheap, it sounds great.

Like mentioned above, good, flat-response studio monitors is the way to go. Mine are Dynaudio BM6 and a Sony amp. I don't use headphones unless I have to.

 

[lemonkid]

Exactly. I used to work for Bell Labs' sister company, Lucent. Even so, I am astonished every day by just how old Shannon-Nyquist Sampling Theorem (1928) is and how simple, elegant and foolproof it remains.

But at Bell labs they did not use a digital, but analog signal. And they did not do this in 1928 but around 1870 or something like that, with analog signals. And because the analog signal gives you an almost infinite bit-rate, only the sample rate counts.
You should also consider that they did these tests with the audio technics of those times. They were done to be used for phone-calls. And they used listeners taken 'from the street'.

The Nyquist-Shannon Theorem is a theory that is used by people who try to explain why it is difficult to hear a difference between a sampled signal and a normal signal. The Theorem does not tell you that it is impossible to hear a difference. In fact it will tell you that it may VERY WELL be possible to hear a difference. Because it will also point out the problem of aliasing.
Modern microphones will 'hear' frequencies higher than the human ear can hear. This will result in distortion of your audio signal. Because the higher frequencies are UNDERSAMPELD.
But all in all I think that good mastering will produce better quality. Provided that the recording was good. And to prevent aliasing recording should really be done at higher sample and bit rates.
see the link:

aliasing

 

[Avatar74]

But at Bell labs they did not use a digital, but analog signal.

This is a vague statement that demonstrates a misunderstanding of discrete time and continuous time sampling, Fourier transform, and other relevant concepts that were built, tested and proven out in practical usage. Specialized mics do exist for applications other than producing sound recordings.

They were done to be used for phone-calls.

I've been working in tech/telecom for 30 years... The bandwidth for phonecalls is around 4 kHz because most of the frequencies required to understand human speech lie between 2-4kHz, so in modern telephony the signal is sampled at 8 kHz (or 2x the Nyquist limit of 4 kHz) and then further compressed using CELP (a form of µ-law compression), to produce intelligible, not flawless audio. That's the test you're speaking of. I'm very familiar with this background because I worked for Lucent Technologies, the sister company to Bell Labs after the spinoff from AT&T.

And nobody here brought up tests Bell Labs did in the 1800s. We're talking about the fact that the math behind the theorem, published in 1928 by Bell Labs engineers, works as predicted in discrete time sampling as proven to this day.

Modern microphones will 'hear' frequencies higher than the human ear can hear. This will result in distortion of your audio signal. Because the higher frequencies are UNDERSAMPELD.
But all in all I think that good mastering will produce better quality. Provided that the recording was good.

No. Modern large-diaphragm condensers used in recording studios, such as my Neumann TLM-103, are designed to roll off at the top of the K-weighted band. Not just that, but, to reiterate, mastering processes and mastering engineers have band limited the signal for quite some time now, which is a key requirement of Nyquist-Shannon explained in the 2019 paper you quoted that doesn't defend any of the points you are making.

I encourage you to watch this primer by Monty Montgomery, demonstrating, among other things, with analog signal generators and analog spectrum analyzers that there is no difference between the original and the reproduction of a band limited signal sampled at 2x the Nyquist limit.

If you're perceiving a difference on any hardware built after 1982, I would immediately suspect a non-flat frequency response from the speakers as the culprit. Barring that, I'd recommend seeing an audiologist.

Some recommended reading:

• Principles of Digital Audio by Ken Pohlmann
• Modern Recording Techniques by David Miles Huber
• Mastering Audio by Bob Katz

 

[lemonkid]

This is a vague statement that demonstrates a misunderstanding of discrete time and continuous time sampling, Fourier transform, and other relevant concepts that were built, tested and proven out in practical usage. Specialized mics do exist for applications other than producing sound recordings.



I work in tech/telecom... The bandwidth for phonecalls using CELP, a form of µ-law compression, is around 4 kHz because most of the frequencies required to understand human speech lie between 2-4kHz, so the signal is sampled at 8 kHz (or 2x the Nyquist limit of 4 kHz), to produce intelligible, not flawless audio. That's the test you're speaking of. I'm very familiar with this background because I worked for Lucent Technologies, the sister company to Bell Labs after the spinoff from AT&T.

And nobody here brought up tests Bell Labs did in the 1800s. We're talking about the fact that the math behind the theorem, published in 1928 by Bell Labs engineers, works as predicted in discrete time sampling as proven to this day.



No. Modern large-diaphragm condensers used in recording studios, such as my Neumann TLM-103, are designed to roll off at the top of the K-weighted band. Not just that, but, to reiterate, mastering processes and mastering engineers have band limited the signal for quite some time now, which is a key requirement of Nyquist-Shannon explained in the 2019 paper you quoted that doesn't defend any of the points you are making.

I encourage you to watch this primer by Monty Montgomery, demonstrating, among other things, with analog signal generators and analog spectrum analyzers that there is no difference between the original and the reproduction of a band limited signal sampled at 2x the Nyquist limit.

If you're perceiving a difference on any hardware built after 1982, I would immediately suspect a non-flat frequency response from the speakers as the culprit. Barring that, I'd recommend seeing an audiologist.

Some recommended reading:

• Principles of Digital Audio by Ken Pohlmann
• Modern Recording Techniques by David Miles Huber
• Mastering Audio by Bob Katz

The reason why they did these tests was not to produce flawless audio, but to learn how many different phone-calls could be sent over a single phone-line at the same time. So they sampled the analog signal to find out at what rate you could still understand each other. I spoke to some of the top engineers at Bell and in no way these tests had anything to do with finding out at what sample rate you could get high quality audio. These tests were only done to find the right sample rate for phone calls.

They did perform High Fidelity tests however in the 1930's. Fletcher did and I have some of his stunning recordings

But you should really try to understand the problem of aliasing. What you do is to view the concept of sampling through the perspective of what the receiver is capable of hearing. But that is not what the Nyquist-Shannon Theorem is about. It is about the SIGNAL that originates from the source. Whatever the principle of digital audio may be, every audio engineer will know about aliasing. No matter how well your microphones will 'roll of'. The signal can only be preserved when you prevent aliasing.

When you read the Nyquist-Shannon Theorem it points out this problem. In no way it says that because the human war cannot hear over 20 kHz you only need to sample at 16 bit 44,1 kHz. On the contrary.

 

[Avatar74]

But you should really try to understand the problem of aliasing.

The "problem" of aliasing doesn't exist in a band limited signal. Period.

every audio engineer will know about aliasing.

I am an audio engineer, for 35 years. I've mastered several professionally produced albums and have several copyrights in sound recordings.

In no way it says that because the human war cannot hear over 20 kHz you only need to sample at 16 bit 44,1 kHz. On the contrary.

Actually that is exactly what it says. As the theorem states:

If a function x(t) contains no frequencies higher than B hertz, then it can be completely determined from its ordinates at a sequence of points spaced less than 1/(2B) seconds apart.

 

[Supermallet]

To answer the original question, at home I use a Schiit Yggdrasil DAC with a Schiit Ragnarok amp, powering both my speakers, NHT SB-3s, and my ZMF Auteurs and ETA headphones.

For portable use I have an Ibasso DC06 Pro dongle that I attach to my phone or iPad and primarily use Sennheiser IE600 IEMs.

 

[nathansz]

Scarlet 8i6 to a pair of kali lp 6 and a pair of big advents

 

[uvafan1]

I use a Dragonfly Black as my portable DAC and Gustard R26 DAC for my home stereo.

 

[1madman1]

I dont really consider myself an audiophile, but I use a Sony UDA-1 w/DSEE enabled at the desktop 95% of the time. For the occasional test that the Sony wont drive well, I use an NI Komplete Audio 6 Mk-II.

I stopped doing portable several years back.

 

[Flynnsworth]

I’m not an audiophile but I run a pair of Neumann KH 80 out of an RME Babyface Pro FS for most of my listening.

For ‘portable’ I use Sennheiser IE-600s through a Chord Mojo 2 DAC.

On the go I use AirPod Pro 2s.

Spotify is my main streaming service as it is unrivaled for music discovery. I listen to my own collection regularly which is CD quality FLAC as a minimum through Roon / HQP, Audirvana or Plexamp. Foobar is nice too (for free).

I have some 24-bit/192kHz stuff from Qobuz and Nugs. It sounds incredible but I wouldn't be able to tell the difference v 16-bit/44.1kHz tracks in blind tests. I've not used Tidal as I don't like the MQA format, but I understand they are slowly replacing it with HiRes FLAC.

The vast majority of people cannot hear a difference between WAVs and tracks compressed at a reasonable bitrate using modern codecs such as Opus or even LAME mp3.

 

[MajorFubar]

It sounds incredible but I wouldn't be able to tell the difference v 16-bit/44.1kHz tracks in blind tests.

Don't worry or feel inadequate: no one can, except some audiophiles who think their flawed hearing and expectation bias rescinds a century-old proven theorem. Audible differences are always caused by differences in the masters.

 

[AustinIllini]

I use a Schiit Bifrost. I like it.