Network Dropouts

[Mike48]

Update: dropouts seem to be gone. I changed three things:

(1) I switched transcoding in MinimStreamer from flac:wav to flac:wav24.
(2) I reboot the MiND (stream receiver) each day before listening.
(3) When I'm done for the day, I exit the control point app, instead of just hitting "stop." This stops the MiND, also.

When I have time, I'll see if I can pin down what matters most. For now, it's easiest just to follow the rituals.

[Mike48]

(21-07-2014 20:07)magister Wrote:  [...]Paul McGowan, the CEO of PSA who is very honest in his discussions on the forum, said that nothing in streaming is easy and that making the new Bridge play gapless reliably even at single-rate DSD has been a difficult engineering challenge.[...]

Clearly, audio is sensitive to network latency in a way that many other applications (e.g. Web browsing) aren't. Still, I imagine it's not the only field in which latency is a problem, and audio has the advantage that sufficient buffering can solve everything, or just about. (In contrast, that might not be the case in some real-time control problems.)

It seems to me that the engineering needed in audio is not terribly exotic, but it's not what audio manufacturers are used to. Big companies like Sony can hire the expertise they need. Some high-end outfits seem to be trying to do it in house, perhaps by trial and error, with bad results for their customers. It is, in the end, the customers who are left holding the bag.

Regarding Paul McGowan's comment on the difficulties of streaming high-rate audio: It seems to me that the extreme data rates starting to be pushed in audio may be overkill, a case of throwing resources at a problem, rather than analyzing it fully. That's my speculation. Still, if PS Audio's new DAC (for example) makes CDs sound as wonderful as they say, why does anyone need quad-rate DSD?

[simoncn]

(26-07-2014 01:55)Mike48 Wrote:  Update: dropouts seem to be gone. I changed three things:

(1) I switched transcoding in MinimStreamer from flac:wav to flac:wav24.
(2) I reboot the MiND (stream receiver) each day before listening.
(3) When I'm done for the day, I exit the control point app, instead of just hitting "stop." This stops the MiND, also.

When I have time, I'll see if I can pin down what matters most. For now, it's easiest just to follow the rituals.

I would be interested to hear what happens if you revert (1) to its previous setting while leaving (2) and (3) unchanged.

[magister]

(26-07-2014 02:10)Mike48 Wrote:  It seems to me that the engineering needed in audio is not terribly exotic, but it's not what audio manufacturers are used to. . . .Some high-end outfits seem to be trying to do it in house, . . .

This, I think, has been part of PS Audio's problem. They make wonderful hardware but it has taken them a while to "get" how to work best with the software end of things. They do seem to be getting better at it. Even so, I take Paul McGowan's comment seriously; if it were just a matter of adding a larger buffer, I think they would have done that early on in the development of the updated Bridge.

Quote:It seems to me that the extreme data rates starting to be pushed in audio may be overkill, . . . Still, if PS Audio's new DAC (for example) makes CDs sound as wonderful as they say, why does anyone need quad-rate DSD?

I think you may well be right about the overkill, although I have never heard quad-rate DSD so I shouldn't make pronouncements. But with PCM, I've always thought that recording at 196 or 176 would be helpful for very complex music (symphony, pipe organ, etc.), yet I have some 24/96 symphonic albums that sound world-class. I guess this all depends on things that I don't know but the engineers do. I can imagine that double-rate DSD might sound better than single, at least in some circumstances, but whether there is any benefit in going further I would question.

And, I may add, that the new PSA DAC is as good as they say. Mine has about 200 hours on it. After only a few hours it was clearly better than my PWD MkIIk, which is one of the best DACs on the market. But the DirectStream just keeps getting better, especially with redbook material. Last night I played a favorite track, ripped from CD, that I hadn't listened to for a few days. I sat up in my chair, amazed to hear a bunch of things I had never gotten before from that song. It is a very special DAC.

[Mike48]

Magister, I'm happy to hear the DS DAC is so good and that it worked out so well for you.

As to what resolution is needed for music, it's been said before, but the best 44.1/16 is far better than the worst 192/24. It seems to me that far more depends on the recording technique than on the resolution, and the practices of recording and mastering engineers are not going to change as easily as adding a few more bits or higher sample rate.

Although the phenomenon is well known, I am still surprised when I listen to a pop recordings with dynamic range of about zero. Unfortunately, those practices are creeping into jazz recording as well. Can opera be far behind?

It's especially sad when young local groups make a CD and it turns out all pumped up and compressed, with no real musical life at all. Those kids aren't fighting for recognition on the airwaves, so there's no reason to hype the recording, but it's just what's done today, and it's what they wind up with.

It seems bizarre that at the same time the equipment industry is pushing high resolution music, the recording industry is working hard to minimize the actual musicality of its products.

[magister]

(27-07-2014 17:32)Mike48 Wrote:  Unfortunately, those practices are creeping into jazz recording as well. Can opera be far behind?
[snip]
It seems bizarre that at the same time the equipment industry is pushing high resolution music, the recording industry is working hard to minimize the actual musicality of its products.

I listen almost exclusively to acoustic music (classical, folk, jazz). So it's distressing to hear that the latter is now getting on the compression bandwagon. And you're right: how ironic to push hi-res, with the potential for hearing much more of what the artists intended, and then squeeze the life out of the artists' creations.

[Mike48]

(26-07-2014 10:32)simoncn Wrote:  

(26-07-2014 01:55)Mike48 Wrote:  Update: dropouts seem to be gone. I changed three things:

(1) I switched transcoding in MinimStreamer from flac:wav to flac:wav24.
(2) I reboot the MiND (stream receiver) each day before listening.
(3) When I'm done for the day, I exit the control point app, instead of just hitting "stop." This stops the MiND, also.

When I have time, I'll see if I can pin down what matters most. For now, it's easiest just to follow the rituals.

I would be interested to hear what happens if you revert (1) to its previous setting while leaving (2) and (3) unchanged.

I tried it briefly this afternoon. No obvious change (i.e., no dropouts). I am beginning to believe that the clicks may not be network dropouts at all, but rather the MiND suffering, e.g., a memory leak or other software (firmware) snafu. That would explain why (2) seems to help.

[Mike48]

More follow up: after changing flac:wav24 to flac:wav and listening to various resolution files intermixed, the clicking began again. I changed it back to flac24, rebooted the MiND, restarted MinimServer, and the clicks went away. I listened for over an hour with no clicks. This leads me to believe that varying bit depths from track to track may have a negative effect on the MiND. I'd need to repeat this several times to know for sure, and I'm not that curious at this point -- I'd rather just listen to music.

[simoncn]

(29-07-2014 07:12)Mike48 Wrote:  More follow up: after changing flac:wav24 to flac:wav and listening to various resolution files intermixed, the clicking began again. I changed it back to flac24, rebooted the MiND, restarted MinimServer, and the clicks went away. I listened for over an hour with no clicks. This leads me to believe that varying bit depths from track to track may have a negative effect on the MiND. I'd need to repeat this several times to know for sure, and I'm not that curious at this point -- I'd rather just listen to music.

This isn't quite conclusive because you rebooted the MiND at the same time as changing the transcoding setting to wav24. It would be interesting to know whether the problem occurs if you play everything at wav24 for an extended period without rebooting the MiND.

If the problem is related to 16-bit WAV streams, it might be caused by the internal processing that the MiND is doing to convert a 16-bit network stream to 24 bits for passing on to the DAC. Alternatively, it might be caused by this conversion within the DAC (if the DAC is receiving a 16-bit stream from the MiND).

[Mike48]

I've been using transcoding at wav24 without booting the MiND for several days. The only time I have heard a click has been between tracks when they have different sampling rates. But we have been traveling, so not listening as much as usual. I'll report again after more time has passed. It will be wonderful if the answer is that simple.