@mrdestructo I guess I'm both an end user and and engineer so was interested in the technical difference. I did an electronics degree which took me into pro audio equipment design for a number of years. In a different role I did quite a bit of research into perceptual audio codecs (which is pretty interesting if you're into this kind of thing).
That 63KI Sig was one of the most musical machines ever made.
I had one of those and it was something of a legend but, yes, DAC technology has come on massively especially in the last ten years.
The best DAC I’ve heard is the dCS Vivaldi. Something else.
I have heard a full DCS stack and really didn't like it; there's no arguing it does a very good job of converting a digital signal to an analogue one but, for my taste (and I know I'm very much not alone here), it doesn't do a very good job of making music; it's very clinical and dry sounding.
My personal preference for DACs are those with tube output stages. Lampizator Golden Gate is currently my DAC of choice.
@Cougar remember that as file sizes go up the prudent listener moves the bit depth above 16 which is where all my files are (24 bit or better). This was what my comment about CD drives was about, remember SACD many moons ago?
The difference is very noticeable, how about you try? I know it works for my ears.
I really enjoy John Darko's youtube channel. I think it's a fair take on the kit, and he does a good job of trying to communicate how one system might sound different to another (which isn't that straightforward a task).
As for DACs, the iFi unit has had some good reviews, so clearly is a good budget pick. Personally, everything sounded like a DAC until I bought a Chord Qutest. Then everything was different. Not cheap but I love mine.
The difference is very noticeable, how about you try?
Nu-huh.
I'd love to. Set it up.
I like Darko's reviews.
Also Andrew Robinson. Z Reviews too (although for ages I was convinced the Zeos was Jack Black!)
Steve Guttenberg's gone downhill since Police Academy though. Really showing his age. 🙂
(j/k)
I've got an iFi Zen, streaming Tidal MQA from USB into a 1970s receiver. Sounds bloody lovely.
At my desk I've got a Dragonfly Black 1.5 into a amp and speakers that are waaaay too big for the desk but whatever, it also all sounds bloody lovely.
I didn't cut corners with the RCA cables or speaker wire.
Typing this listening to Tidal HiFi from my MacBookPro through DacMagic 100 then Class D Sonos Amp and Rogers JR149's on isoAcoustics stands (a bit big on the desk). I was going to go for LS3/5a's, but 1) cost and 2) tonality for music. The stands made a huge difference on the desk. I tested USB from work laptop, optical from MBP and SONOS stream. I was considering refurbishing the Quad 303 in place of the SONOS Amp, which doesn't really get used for streaming in the office!
@judetheobscure My ears are reasonably good but, not good enough to need more than 24 bit files in the car. I do play the phone through a plug-in DAC and wired headphones as that's a nice experience too, I don't have any kit that would cope with DSD above 128 and no funds to purchase any (we're into bike S-1 territory if I do!).
@Cougar I'm unable to empirically prove to you that my ears can tell the difference, because:
1. They're mine.
2. I'm unable to share the processing that my ears use with you in a non destructive manner.
3. Even if I do set you up a testing environment you will shortly be unable to attend as our "Boris Christmas Lockdown II - Lockdown Harder" is due to start.
If you can't tell then all well and good, you can't tell.
How many times do I have to write this? Music on a CD is data not audio. A 1 on a CD is exactly the same as a 1 on your hard drive. Exactly. A CD is not vinyl, there are no 0.5s here. A cheap CD drive won’t result in inferior audio reproduction, the data is either intact or it isn’t.
Except its not quite as simple as that, you may not get a 0.5 but you can get an error that means a 1 gets read as a 0 and vice versa. Also a CD drive is treated like a block device on a PC so you end up reading the data in sectors, red book formatted CD's are not designed to be read like that and thus don't have the framing to allow for accurate positioning on the disk, so your sector reads can overlap or you can end up with gaps between them. So you end up having to read multiple sectors and overlapping them and doing some shifting along to make sure you don't get and gaps etc. So ripping a CD does not always end up with an exact copy. So the better the transport on the CD the more chance you have of not having bit errors, the less bit errors the more chance you have of correcting them. Bit errors can cause noise in the final audio, eventually you will get skips and obviously audible artifacts, but before that there will be general distortion of the audio signal.
I did a final year project at university that involved some ripping of audio CD's and storing them on DAT tape. I discovered it was not as simple as I initially thought.
What @bazzer says +1
Many CD ripper devices have some sort of feature built in to try and make sure the rip is 100% accurate. Some will read the CD multiple times and compare results. Some will generate a checksum based on the CD data read and compare with online database of checksums for the same CD etc.
When you play a music CD on a CD player the data it reads is not 100% accurate to what is on the CD. The CD player will be applying some error correction and filtering to fill in the blanks.
Similar thing occurs in the SPDIF connection, its not guaranteed to transmit with 100% accuracy and the DAC will be applying error correction and filtration to the signal it receives.
I've only done a quick test of High Res music, I downloaded a 24-bit copy of album in WAV format and compared that to 16-bit WAV file ripped bit perfect from the CD. I couldn't hear the difference. So just stuck with 16-bit copies. Been meaning to do a more in-depth test. Don't think my server or DAC support DSD formats? Will have a deeper looking to that.
Apple lossless is a bit of a contentious subject with audio fans.. you could install Roon on your Mac and then send it to a Wi-Fi streamer connected to a usb DAC or to a combined streamer / dac (there are many) but Roon only supports Qobuz and Tidal. You could send it via AirPlay but that’s limited to CD resolution, you could use Chromecast but again CD and not Apple Music.. you could attach a DAC via USB (iFi Zen etc) it be aware hat Mac usb drivers won’t support the full resolution of hi-res (Mac uses Core Audio drivers only (Apple deleted ASIO support from their OS with the release of OSX), there is no ASIO support on Mac.
Unfortunately macOS does not understand DSD. That’s why the DSD over PCM standard has been defined: to send native DSD data inside what macOS believes to be a PCM stream.
But this has a drawback: as it needs to embed markers for the DAC to recognize it as DSD, the needed bandwidth is doubled. Thus on macOS, only half of the DAC max DSD rate is available.
This is why Apple Music “limits” the resolution as they know their hardware can’t support it…. it’s lossless as long as it wasn’t too high res to start 😀.
Basically you need to separate High Res and lossless (which is just a marketing term).
So long story short buy a Bluesound Network streamer Dac connect this via RCA to your AMP and sign up for Tidal (or Quobuz) .. the stream comes straight from the servers via your network to the Node and your Mac (or phone, iPad or whatever) acts as a remote control for this stream
Hope that helps.
I’m unable to empirically prove to you that my ears can tell the difference
Sure you can. You don't need to prove it to me, you just need to prove it to you. I'll believe you.
Get your partner to cue up a few songs whilst you aren't watching. Can you reliably tell the difference?
Except its not quite as simple as that
Sure it is. Because...
The CD player will be applying some error correction and filtering to fill in the blanks.
... that's how ECC works, see that word "correction" right there? If there's errors, it will interpolate the missing data. If errors exceed the capability of ECC then you ain't going to get 'noise' but rather quite the opposite.
I did a final year project at university that involved some ripping of audio CD’s
How long ago? Were they ripped down a digital data bus or over an analogue CD-ROM audio cable?
not good enough to need more than 24 bit files in the car
Yeah pretty pointless unless you have a soundproofed car and hear zero tyre / wind noise when you're listening.
DAC's do colour the sound and some do sound "better" than others.
There was a site that had an extensive list of DACs, reviewed and charted, but I can't for the life of me recall the site address. TBH though it's like anything. Subjective opinion is entirely YMMV. I do love when I see "x DAC blows Y DAC out of the water!" as it's impossible to have identical listening setups and ears.
Get this DAC it's the best I've ever heard!
Big Spotify fan and eagerly awaiting the hifi tier if/when it even happens.
Ripped using a CD-ROM drive completely digitally. I thought it would work in the same way as reading data from from a data CD-ROM its not. If you ask for logical block you don't always get the block you ask for as the disk can't position properly as the format is not designed to do that, its not needed when streaming from a device.
Also the ECC can't correct everything and the odd bit error will not cause a catastrophic loss of audio, in fact you'd probably not even notice if it was the least significant bit.
This is why modern ripping software does a lot of clever stuff to make sure it gets the best copy possible.
Edited for more detail
Data is stored in 2352 block 2048 data the rest ecc, the problem is you read 2352 bytes and due to the problems outlined it could be shifted by a few bytes so you might get a few bytes from the previous block and not all the data from the block you wan't. So you end up having to do overlapped reads of multiple blocks and sliding them along against each other to match up data. This is all well and good but if one read has a bit error (remember this is pre ecc as the ecc is in the data still) then you can't match it up as easily.
Also due to the fact its streamed off the disk when playing if ecc fails then it can't just reposition and do a re-read like a disk drive or a CD-ROM does. So you might end up with bit errors in the stream, this could be anything from inaudible to big clicks. When it skips or fails to play that's normally when its so bad it can't even track the data on the disk.
So a better transport can give less bit errors, and we have not even got to the effects of the clock and its effect on frequency domain distortion.
So if you were to play a CD and record the digital out as data, it would not be the same every time. You would get un-correctable bit errors in different places.
Yet again the forum delivers! Someone whose field of expertise is in CD drive technology appears to explain things in a way we all can understand. Thanks @bazzer
Edited for more detail
Data is stored in 2352 block 2048 data the rest ecc, the problem is you read 2352 bytes and due to the problems outlined it could be shifted by a few bytes so you might get a few bytes from the previous block and not all the data from the block you wan’t. So you end up having to do overlapped reads of multiple blocks and sliding them along against each other to match up data. This is all well and good but if one read has a bit error (remember this is pre ecc as the ecc is in the data still) then you can’t match it up as easily.
Also due to the fact its streamed off the disk when playing if ecc fails then it can’t just reposition and do a re-read like a disk drive or a CD-ROM does. So you might end up with bit errors in the stream, this could be anything from inaudible to big clicks. When it skips or fails to play that’s normally when its so bad it can’t even track the data on the disk.
So a better transport can give less bit errors, and we have not even got to the effects of the clock and its effect on frequency domain distortion.
So if you were to play a CD and record the digital out as data, it would not be the same every time. You would get un-correctable bit errors in different places
STW at its best.
Well, that's me told. (-:
Interesting, I'll go do some reading. Thanks.
@cougar in fairness, when I started I assumed it would work how you envisioned it working and it would be perfect as its digital. However the subtle bit is that if you get multiple bit errors then the ECC can't correct (I forget how many it can correct) and it can't do a retry when playing a CD so those errors are now in the bitstream.
It turns out it's not quite that simple in reality.
@bazzer very interesting insight and not one I’m remotely challenging as I have zero technical knowledge on this subject (I just use my ears), but in terms of generating a 'bit perfect' copy of a CD stored as a digital file, is the variability introduced by the CD-ROM drive negated by the processes undertaken by the ripping software (in the checking and correcting of that copy)?
so those errors are now in the bitstream.
Right. Whereas the conclusion I reached was that those errors would cause the rip to go "it's bollocksed" and dump the process.
I'll go back and refresh my knowledge at some point, I haven't looked at Coloured Books since the 90s. I'm labouring under the notion that Audio CD is held to the same standards as CD-ROM and that's blatantly not the case if you stop and think about it for 30 seconds (which I didn't). Audio CD playback doesn't need to be bit-perfect, does it. Does reproduction from a CD-ROM drive have different connotations from playback from a hi-fi device I wonder?
Hmm. Strange things are afoot at the Circle K.
To a point yes, ripping software can re-read data something that a CD-Player can't do, however with all the caveats about positioning etc. However if it continuously reads data again and its got bit errors ecc can't correct (damage/fingerprint or even a poorly aligned/focused laser etc etc) it won't get a perfect copy of what the master digital media contained.
For the most part this is very subtle and it works just fine, I would not worry about it 🙂 Obviously once you have the PCM data ripped to your computer you can copy it without degradation.
@cougar I was talking about a CD-Player when I was talking about the bitstream really.
Years ago a lot of ripping software didn't even bother to do the ECC it made it too slow, I am sure modern stuff might if you ask it to rip in the best quality. However once you have an error you can't correct your only option in that case is to abort the rip. So it's all a compromise.
If I remember correctly its a hamming code style ECC with a 1 to 4 bit redundant vs non redundant data. I would have to check how many bit errors that can correct and how many it can detect. However due to the very nature and size of blemishes etc on disks you tend to get bursts of errors rather than random.
@bazzer I try not to refer to folk as experts as it has awful flashbacks to noughties training where one was a drip under pressure and someone here will also remember it and go down the dad-joke route.
An impressive depth of knowledge. Chapeau.
Do you need to use a laptop/Mac?
I have the Apple lossless library subscription thingy.
I use a Yamaha WXC-50 and use my phone to stream directly to it from the internet.
I have it going straight into a pair of fully active studio monitors and am willing to bet money the sound is far far better than traditional hifi costing thousands.
Anyway, once set up properly the Yamaha is amazing.. just a thought
I have it going straight into a pair of fully active studio monitors and am willing to bet money the sound is far far better than traditional hifi costing thousands.
Active speakers are traditional hifi aren't they? Although it is interesting that active drive is not more common in speaker design; it is something of a flaw I think in the prevailing fashion. I read an interview with the CEO of a company called Stenheim, Swiss manufacturer of high end speakers, who said they would like to produce all their speakers in active configuration but the market demand for this approach is just not high enough to justify bringing those designs to market.
Active speakers are traditional hifi aren’t they?
I’m not really sure what you mean. Fully Active speakers use a crossover split either digitally or at line level or below, with the signal fed to discrete amplifiers designed to feed either a tweeter or a woofer each.
Traditional hifi throws it all together and splits the signal with a passive crossover at 75+v which results in all sorts of horribly broad filters and phase disruption..
(and yep, it is interesting....the pro audio world have used active monitoring for about 40 years (it will always be more accurate, is a "better" design and does away with a lot of problems inherent to passive systems).
However, the hi-fi world remains mostly passive because it's driven by the marketing of an "upgrade path". People want to save up for the next best thing, the next shiny thing etc. As a consumer I understand that but as an ex-engineer (sort of) I find it excruciating when magazines lay into Active designs purely based on it not fitting their business models/advertising streams very well)
I generally stay well away from conversations like this as there'll always be someone arguing against the engineering advantages of active systems as they don't want to feel like their £4000 system isn't as accurate as £800 monitors... 😀
(and in the interest of balance, my monitors sound fantastic but are a bit ugly as most are)
Funny you should say that.
I was considering replacing the iMac + AE with one of these.
https://www.richersounds.com/yamaha-wxad10-darkgrey.html?nosto=nosto-page-category2
Traditional hifi throws it all together and splits the signal with a passive crossover at 75+v which results in all sorts of horribly broad filters and phase disruption..
Well yes, but there are lots of active speaker designs in the audiophille world. ATC spring to mind as a leading proponent. Audio Note (UK)'s top end speakers all have external crossovers (though I'm not sure if they are active or passive in execution).
I'm not sure why the fashion is for an internal passive crossover but I'm not sure it's about the upgrade path; external active crossovers don't preclude this and in fact they increase the potential for endless updagrades exponentially (because now you can even more boxes). Naim are big into this for instance (active external crossovers).
Also, whilst the thing that audio engineers care about is indeed 'accuracy', what most audiophiles care about is 'music'. There's a subtle difference between the two but when the two intersect then you get an especially good result.
Once you go down the streamer route, then you have to factor in the control/app part (unless you have Roon). I have a Yamaha streaming amp in the garage but the MusicCast app isn't very good, whereas the Bluesound streamer/active Flex speakers in the main house uses the BluOS app and is so much better. The upshot is I hardly ever use the Yammy to stream stuff, whereas the Bluesound stuff gets used pretty much ever day.
Thanks for all the input people, much appreciated.
I have decided against a non Apple streamer and will be giving a separate DAC a go.
I have a Cambridge Audio DAC Magic 100 arriving tomorrow.
Updates to follow.