[Kamil21]

I suspect that much of the difference is down to the distance that the signal travels. For CD it is as short as it can possibly be. For the Melco, it is the length of a 1m USB cable.

It follows that if the Blu2 transport were properly designed, the likelihood of an external source sounding better at 44.1k would be remote. At best, they would sound the same.

 

[DaveRedRef-III]

Well, I've got today off so I have done some more comparisons this afternoon - thankyou Attorney, but you opened a can of worms that I had been resisting! I have plugged the Melco directly into the USB input of the Blu II and compared that to CD playback and the results are as before, but less pronounced. CD rips sound even closer to the CD and highres files sound better as well. To put it simply, the CD just has greater transparency, delicacy and finesse to the sound but I must stress that to a casual or even a fairly focussed listener, these differences would be very small and maybe not even distinguishable at all to some people. If you distinguish no difference, I'd say that you should be thankful for that!

I suspect that much of the difference is down to the distance that the signal travels. For CD it is as short as it can possibly be. For the Melco, it is the length of a 1m USB cable. For the mRendu, my PC has to call the files from the Melco, via the router, playback via Roon back through the router to the mRendu and then through a very short USB cable into Blu II. This is quite a journey and I think, to get the most from file playback, I may need to get a Roon server that sits in my rack and either feeds directly into Blu II or out to mRendu or similar and on into the Blu II, so that the signal avoids a two way journey through the router. I've looked at this but, as Attourney says, it is complex, potentially expensive and seems to be an area that is in a state of flux at the moment so I am also going to wait and see.

The main thing though is that music through the BluDave is a real joy to listen to and can be emotionally moving, surprisingly so in fact at times.

I am sure, given time, Rob will design a suitable streaming solution Malc.

Thanks for the feedback and for taking the time to investigate. Fwiw I have always believed CD had significantly more to offer. Which is why I didn't jump to file streaming. I noticed early on the reviewers never compared top-dollar streaming with top-dollar CD transports. The silence was deafening.

 

[ecwl]

I don't know why I tortured myself by listening to the same track over and over again (and another) on Blu 2 off my Focal Utopia headphones to find answers for myself as to whether the CD transport sounds better than USB. I ended up choosing Hamilton's It's Quiet Uptown.

At first, I was surprised CD transport sounded slightly better, very slightly. There seems to be slightly more dynamic transient, slightly more air, slightly more dimensionality to the vocals (hard to hear off headphones) but it also sounded less smooth than my PC.

And then, this is where things got interesting. I normally run my PC headless and control it with JRemote on my iOS devices. I had to use Microsoft Remote Desktop to log into the system to make further adjustments. Even though the track is 16-bit 44kHz, by the very nature of the ASIO driver, 32-bit 44kHz is being sent to Blu 2. So I switched over the WASAPI driver and at 32-bit 44kHz which is the automatic default setting, I hear no difference. But when I switched to 16-bit 44kHz on the WASAPI driver, a small amount of the dynamic transients that I was missing from CD transport playback returned. And then I realized that JRiver normally has TPDF dithering automatically turned on even in bit-perfect playback mode. So I turned off TPDF dithering and then a small amount of the dynamic transients came back with just a slight decrease in smoothness to the sound. The previously slight difference between CD and USB was still present but it's definitely marginally smaller. But I would say it's still a consistently audible difference.

But I almost never have my laptop constantly logged into the PC with Remote Desktop to listen to music. So when I logged off my PC and then remotely controlled the headless PC like I normally do with JRemote. Now, I'm no longer certain whether I hear a difference between the CD transport and the USB output. The dynamic transients of the piano almost sounds identical now. The dimensionality of the vocals were also almost identical. I wonder if the CD has more air and the PC USB sounded slightly smoother. Not totally sure. But I find that I was having brain fatigue.

So I tried switching to Grieg's Peer Gynt Suites No. 1 4th movement/part ("in the Hall of the Mountain King") conducted by Karajan. Maybe CD transport soundstage was better and slightly more sorted out. I listened to the track off the USB and then CD then USB. I have to admit, I'm just not sure by the time I re-listened to it via USB. But I don't hear a difference in terms of dynamic transients, instrumental timbre, air or even smoothness of the sound. That's when I need to mentally take a break from this.

It was a very interesting exercise but not necessarily an enjoyable one. Everything sounds phenomenally great either way. Even in the worst sounding playback via ASIO 32-bit 44kHz with TPDF dithering from JRiver, it just sounded great to begin with. I had to listen hard to hear the CD transport improvement. Maybe I'm getting old.

The other interesting thought/possible discovery is that sometimes when we think about bit-perfect playback off computer audio, we may actually not be getting it because our streamer might automatically default to 24-bit 44kHz or 32-bit 44kHz due to the USB driver. And my experience with other streamers is that they always display 16-bit 44kHz bit-perfect playback but deep down we really don't know whether extra 0's are added to make the signal 24-bit 44kHz for transmission purposes. And maybe Blu2 handles the two signals differently? Maybe not. Not sure. There was an audible difference in my system between 32-bit 44kHz and 16-bit 44kHz output via USB.

Ultimately, I may not be the best person to trust on this because I'm biased. I've optimized my stereo setup in my living room but I need subtle parametric equalization to get rid of that last bit of bass peaks that I couldn't with a physical setup. So I have to DSP my music playback unfortunately which is suboptimal from a digital standpoint. So even if CD transport playback sounds marginally better in my system, I simply would not be able to effectively take advantage of it without turning my living room into a female-unfriendly stereophile shop. So back to listening to Peer Gynt Suites off my speakers which sound simply stunning.

 

[Malcyg]

Ha, yes it is quite tiring and certainly, I wouldn't call it enjoyable. I usually only do it when I am contemplating buying something!

Your observations seem pretty much in line with mine I thought ecwl. I find it easier to spot differences by playing the same track and switching between sources so that I can compare the same segments from each source. As you do it and focus on the sound, the differences to start to become progressively more prominent. It's like a ticking clock which seems to get louder and louder the longer you go and can't get to sleep!

 

[Jawed]

Does Blu 2 use a full set of 1 million samples for upsampling music files of higher sample rates than 48KHz?

It seems like the elapsed time used by the upsampler (it's about 1.3 seconds of samples for CD at 44.1KHz using Blu 2) is a critical factor in perceived quality. If true then a 96KHz file upsampled to 768KHz will also require 1 million taps, since the elapsed time is determined by the upsample rate, 768KHz.

So, if 96KHz music is upsampled with less taps, and it's true that elapsed time is a critical aspect of the sound quality, then it might be reasonable to assume that Blu 2 is solely optimal for 44.1 and 48KHz music sources.

Now playing: Swans - Why are we Alive?

 

[Malcyg]

I am sure, given time, Rob will design a suitable streaming solution Malc.

Thanks for the feedback and for taking the time to investigate. Fwiw I have always believed CD had significantly more to offer. Which is why I didn't jump to file streaming. I noticed early on the reviewers never compared top-dollar streaming with top-dollar CD transports. The silence was deafening.

Well, perfection for me would have been M Scaler inside the DSX1000 chassis with Ethernet and USB input. I have ditched CD once and would certainly be happy to do the same again. That said, I am finding that I am starting to quite enjoy the experience of popping a CD and having no iThings in my hand, just the Dave remote which controls both Dave and Blu. It is a novel and quite relaxing experience.

The biggest point for me about BluDave though is the fact that I won't be buying any highres files any more. It has only taken 37 years, but maybe CD finally delivers on its initial promise.

 

[Malcyg]

Does Blu 2 use a full set of 1 million samples for upsampling music files of higher sample rates than 48KHz?

It seems like the elapsed time used by the upsampler (it's about 1.3 seconds of samples for CD at 44.1KHz using Blu 2) is a critical factor in perceived quality. If true then a 96KHz file upsampled to 768KHz will also require 1 million taps, since the elapsed time is determined by the upsample rate, 768KHz.

So, if 96KHz music is upsampled with less taps, and it's true that elapsed time is a critical aspect of the sound quality, then it might be reasonable to assume that Blu 2 is solely optimal for 44.1 and 48KHz music sources.

Now playing: Swans - Why are we Alive?

I did question whether M Scaler is optimised for 44.1 - it would seem logical given that it is primarily a CD transport. It certainly sounds like it has been optimised to redbook to me.

 

[ecwl]

Of course, now that I think about it some more, I realized after parametric EQ, there may actually be no need for me to add TPDF dithering in JRiver. After all, Blu 2 already has noise shaping. And then after reducing the volume digitally in DAVE, there is another layer of noise shaping. So I'm not entirely sure if TPDF dithering is really necessary post-truncation from the parametric EQ to address the quantization error and distortion from JRiver because Blu2/DAVE might take care of that. I feel like I'm hearing more clarity and transparency with a slight improvement in dynamic transients on playback without any detriment to the sound so far. And unlike the previous hour of A/B listening, the difference was easily audible. I'll assume it's an improvement until someone who understands this more than I do tells me otherwise.

 

[stvc]

Jul ... Anyone got their Blue MII recently?

 

[ray-dude]

My dealer is saying August for wave 2 to US

 

[Deftone]

Finally, I may start playing CD a bit more now. File playback and streaming will still be my main approach but, if I want to sit and listen properly to an album, I shall now be more inclined to use the CD with BluDave. The downside is that it seems that I may still have work to do on the file playback side as well as evaluating whether redbook rips sound better than the highres files that I have purchased - I hope not, but suspect that they may.

I have been skeptical about hires 24/96+ from the start, i am glad that i have stuck with CD and the only time i have ever purchased hi res material (twice) is because of the better dynamic range over the CD version.

 

[Sonic77]

Jul ... Anyone got their Blue MII recently?

Nope, who knows when, starting to think who cares at this point.

 

[Rob Watts]

I dislike doing serious A/B comparisons as I find them to be quite tiresome, so I only generally do them when faced with the prospect of spending some money! However, given my observations in post 872 above, I thought that I should test out the Blu II CD and compare that with hires file playback because my initial reaction to BluDave was that hires downloads may well be unnecessary for me moving forwards. The A/B results confirmed my initial impressions, but were surprising in one respect.

First, I tested some 96/24 PCM hires downloads vs the same albums on CD. Both file and CD were played through the BluDave, file via USB, and I switched between the two to compare. Whilst the hires download was initially impressive, further listening and comparison actually revealed it to be slightly compressed, congested and bloated by comparison to the CD - slightly overstated and trying too hard to impress. The CD by comparison had greater subtlety and finesse, was more transparent and 3 dimensional, more airy, open and dynamic, less fatiguing and just generally more natural, relaxed and 'realistic'. The more I listened, the greater these differences became until they were quite substantial in my mind, but that's how it generally goes with focussed listening.

This made me think that I should test playback of the ripped album file against playback of the CD from which the file was ripped. This is where the surprise came in because whilst CD replay was again superior, the difference was less than it had been between a CD and a highres download. This could be because it is the rip from the same CD, so it is more of a like for like comparison and therefore rules out a number of potential variables, or it could be that with Blu II being a CD player, Rob's code for the MScaler is optimised for red book resolution. Maybe upsampling higher res files gives it that slightly bloated and overblown sound. Whatever, it is a bonus for me as it means that I can continue to buy CD's and rip them instead of highres downloads which are expensive by comparison.

Finally, I may start playing CD a bit more now. File playback and streaming will still be my main approach but, if I want to sit and listen properly to an album, I shall now be more inclined to use the CD with BluDave. The downside is that it seems that I may still have work to do on the file playback side as well as evaluating whether redbook rips sound better than the highres files that I have purchased - I hope not, but suspect that they may.

An interesting event happened when I was in Tokyo for the International Audio Fair which is pertinent to this post. Hiroko, Chord's high-end distributor in Japan, took me to a demo supposedly showing the "superiority" of HD files, as they were using Dave to do this. The demo was conducted in Japanese; they first played the 44.1/16 version, and it sounded superb; transparent, deep sound-stage, and emotionally involving. They then played the HD version (I recall it as 176.4/24 but was definitely PCM), and the demonstrators started grinning like Cheshire cats; but the sound was much worse and very similar to your experience - it sounded unnaturally soft, warm and out of focus; and the sound-stage - which before was deep and very layered - now sounded flat with no real depth perceptive. I was shocked how much worse it was, and initially very puzzled. Then the demonstrators started talking more, and then it clicked as the word DSD kept cropping up.

Afterwards I asked Hiroko what was going on, and she said that the first CD was the original from the '90's; and the new HD recording was based on a DSD re-master from the original, then converted to HD PCM. And soft, over warm and bloated with poor depth is exactly what DSD sounds like; and the technical reasons why the process sounds the way it does I fully understand - there are very sound technical reasons for the over warm and flatter sound-stage of DSD.

So I am sure what you are listening too is not the same master, so this is not a pure test of CD against HD, as your observations mirrors my experience in this demo. I though have obtained two files where the 44.1/16 is obtained from the master - so can hear absolute difference from 44.1/16 against 96/24 - and the 96/24 (the original master) sounds very much better - deeper sound-stage, more transparency - than the CD version. Note that the higher the SR the more important it is to engage the HF filter. Indeed, I always purchase modern recordings as HD if I can be sure the recording is the original master.

As to whether the WTA filter being optimized for CD - actually this is not the case, as it is the identical filter with identical coefficients, and identical amount of stored data samples, and with the DSP cores running at exactly the same number of instructions per second. All that happens (for single speed 44.1 against double speed 88.2) the over-sample ratio is from 16 to 8 and half of the coefficients are not used (as the over-sample ratio is halved). The filter performance and character is identical (so if it was -3 dB at say 22.000 kHz it would now be -3 dB at 44.000 kHz for double speed).

So the rule of thumb is simple; you can't make meaningful observations of anything unless you know the exact provenance of your source or test.

 

[Malcyg]

An interesting event happened when I was in Tokyo for the International Audio Fair which is pertinent to this post. Hiroko, Chord's high-end distributor in Japan, took me to a demo supposedly showing the "superiority" of HD files, as they were using Dave to do this. The demo was conducted in Japanese; they first played the 44.1/16 version, and it sounded superb; transparent, deep sound-stage, and emotionally involving. They then played the HD version (I recall it as 176.4/24 but was definitely PCM), and the demonstrators started grinning like Cheshire cats; but the sound was much worse and very similar to your experience - it sounded unnaturally soft, warm and out of focus; and the sound-stage - which before was deep and very layered - now sounded flat with no real depth perceptive. I was shocked how much worse it was, and initially very puzzled. Then the demonstrators started talking more, and then it clicked as the word DSD kept cropping up.

Afterwards I asked Hiroko what was going on, and she said that the first CD was the original from the '90's; and the new HD recording was based on a DSD re-master from the original, then converted to HD PCM. And soft, over warm and bloated with poor depth is exactly what DSD sounds like; and the technical reasons why the process sounds the way it does I fully understand - there are very sound technical reasons for the over warm and flatter sound-stage of DSD.

So I am sure what you are listening too is not the same master, so this is not a pure test of CD against HD, as your observations mirrors my experience in this demo. I though have obtained two files where the 44.1/16 is obtained from the master - so can hear absolute difference from 44.1/16 against 96/24 - and the 96/24 (the original master) sounds very much better - deeper sound-stage, more transparency - than the CD version. Note that the higher the SR the more important it is to engage the HF filter. Indeed, I always purchase modern recordings as HD if I can be sure the recording is the original master.

As to whether the WTA filter being optimized for CD - actually this is not the case, as it is the identical filter with identical coefficients, and identical amount of stored data samples, and with the DSP cores running at exactly the same number of instructions per second. All that happens (for single speed 44.1 against double speed 88.2) the over-sample ratio is from 16 to 8 and half of the coefficients are not used (as the over-sample ratio is halved). The filter performance and character is identical (so if it was -3 dB at say 22.000 kHz it would now be -3 dB at 44.000 kHz for double speed).

So the rule of thumb is simple; you can't make meaningful observations of anything unless you know the exact provenance of your source or test.

Thank Rob, that is interesting. I did wonder about the sources being different and other possible variables which is why I then tested some files against the CD that they were ripped from so that I knew I was testing like with like. From your comments then, even better sound may be possible from a high res file as long as it is a good high res file.

I have read about hires files being upsampled and questions as to what you are buying so, for me, I think I shall stick with CD from now on. It is cheaper, provides a hard back up once ripped, I know where the file came from and I have the CD to play if and when I choose.

 

[Kamil21]

An interesting event happened when I was in Tokyo for the International Audio Fair which is pertinent to this post. Hiroko, Chord's high-end distributor in Japan, took me to a demo supposedly showing the "superiority" of HD files, as they were using Dave to do this. The demo was conducted in Japanese; they first played the 44.1/16 version, and it sounded superb; transparent, deep sound-stage, and emotionally involving. They then played the HD version (I recall it as 176.4/24 but was definitely PCM), and the demonstrators started grinning like Cheshire cats; but the sound was much worse and very similar to your experience - it sounded unnaturally soft, warm and out of focus; and the sound-stage - which before was deep and very layered - now sounded flat with no real depth perceptive. I was shocked how much worse it was, and initially very puzzled. Then the demonstrators started talking more, and then it clicked as the word DSD kept cropping up.

Afterwards I asked Hiroko what was going on, and she said that the first CD was the original from the '90's; and the new HD recording was based on a DSD re-master from the original, then converted to HD PCM. And soft, over warm and bloated with poor depth is exactly what DSD sounds like; and the technical reasons why the process sounds the way it does I fully understand - there are very sound technical reasons for the over warm and flatter sound-stage of DSD.

So I am sure what you are listening too is not the same master, so this is not a pure test of CD against HD, as your observations mirrors my experience in this demo. I though have obtained two files where the 44.1/16 is obtained from the master - so can hear absolute difference from 44.1/16 against 96/24 - and the 96/24 (the original master) sounds very much better - deeper sound-stage, more transparency - than the CD version. Note that the higher the SR the more important it is to engage the HF filter. Indeed, I always purchase modern recordings as HD if I can be sure the recording is the original master.

As to whether the WTA filter being optimized for CD - actually this is not the case, as it is the identical filter with identical coefficients, and identical amount of stored data samples, and with the DSP cores running at exactly the same number of instructions per second. All that happens (for single speed 44.1 against double speed 88.2) the over-sample ratio is from 16 to 8 and half of the coefficients are not used (as the over-sample ratio is halved). The filter performance and character is identical (so if it was -3 dB at say 22.000 kHz it would now be -3 dB at 44.000 kHz for double speed).

So the rule of thumb is simple; you can't make meaningful observations of anything unless you know the exact provenance of your source or test.

Wow, this could nullify many 'sound impressions ' on this forum . I wonder how many recordings, CD or otherwise that are 'damaged' by the upsampling/downsampling process end up being used as reference recordings. Could it be that there are more CDs out there that are 'original' than there are coming out of streaming or remastered downloads.

If so, this is the best argument I have heard so far for getting the MScaler/CD transport such as Blu2.

We should now be looking out for Studio Master-Bit Perfect labelling!