Streamers in 2023 - Discussion, Impressions, etc

Negative results can also be interesting.

Back in the day when people hated AirPort Express (this is pre-Bifrost for me), I could not hear a significant jump between a USB stick and the optical output in the express going into whatever DAC I had. Turns out they were equally subpar.

Straight out of a Mac or the Nucleus via USB with Unison input works pretty reliably for me. I use 2 Pi2AES, one with AES Coax into Yggdrasil and another one via SPDIF into Gunginr. Cable runs are short. I hear nothing different there. I also use a SPDIF out of a CXNv2. No concerns. It may be that both streamers are just indistinguishable from each other. I sometimes use a Pro Ject streamer (tiny one) to test stuff, but for all practical purposes, it’s a Pi. Can’t say one is warmer than the other.

If people have concrete experiences of sound differences I’m definitely interested, because I may just be missing something to obsess about. For me the jumps, if any, are about which input to the DAC I’m using.

 

Streamers hardly existed before the Pi, there was Logitech and DNLA, and some fancy cats doing UPnP streams, but UPnP was used for multi-room device...one player, multi room thing, like Sonos. Logitech was playing all your files from one computer just on another device on another room. This was all before actual internet streaming services.

But Pi devices had a bunch of noise and jitter on the line that was sometimes worse than just using USB on your computer, but the convenience of a separate headless device doing ONE thing is quite appealing, but you still have to keep the noise and jitter down.

To me, a source (USB receiver, streamer, transport) should not play a factor in synergy. It should just BE the source. Sure, some older USB devices can sound warmer or rounder, but that should not be way to tame a bright system for example.

Also, everyone's needs and preferences are different in how they want to interact or control their streamer. So I think this is where the emotions come in.

 

Sorta. The Squeezebox (I forget whether this is before or after Logitech bought them) needed Squeezebox Server/LMS, but it could also integrate with the Rhapsody subscription streaming service, in basically the same way that Roon needs a server but can work with Qobuz/Tidal -- there was a cloud component that acted as a middleman.

 

Oh yeah, I forgot about Rhapsody and early "radio" style streamers like and Pandora. But yes, Squeezebox can also connect to internet radio streams. But the "server" was already your home computer where you had all your sound files anyway. Then you could use the Logitech device to play them via DNLA. At least this is how I remember it with a friend who had an early Logitech box and could play files from his computer in his garage workspace.

 

There was the radio-style thing, but they also had an on-demand subscription service that's very recognizably like modern streaming services, as early as 2007. Using that with the Squeezebox was as seamless as modern Roon+Qobuz/Tidal. (And technically it wasn't DLNA, it was SlimProto, which was both more and less sophisticated than DLNA.)

I'm using the past tense, but of course there are people in modernity who still run this stuff.

 

A more philosophical question, I suppose. Do you think we have gained anything from the days of the Squeezebox? I'm not sure the sound quality has improved. Unless you are invested in hi-res or MQA. I suppose the biggest convenience is running an RPi or other streaming digital device allow us to have an always on, always accessible access to our music.

Perhaps more related to the thread, once you get to the Pi2AES level, can the improvements in jitter be heard and therefore worth more money? And as Michael Kelly has asked before, below a threshold where it interferes with reading the digits, is noise important in a digital streamer?

I don't have the answers here, but I am curious.

 

Just remember jitter is not noise per se. Jitter affects the time correlation of the music (soundstage, depth, etc). This can be quite noticeable even at low levels. And as you say, noise on the digital line is not an issue unless it cause the bits to be unreadable.

 

I am a frequent lurker but have been inspired to add some impressions of pi2AES vs Zen stream. I don't know if this is the best place for this, but I blame what follows on this thread. While it was never my intention to compare streamers, I ended up with 2 streamers and overkill power supplies.

My chain is Jriver >zen stream (illuminati power supply) or Pi2AES (Farad PS)>spdif cable>Bifrost 2OG> Freya S passive> rebuilt vintage onkyo integrated amplifier (being used only as power amp via power amp input)>triangle Antal 40th anniversary> my non-magical ears.

1. Music : Alone Together- Chet baker
- Zen Stream: blacker background. Soundstage appears deeper.
- Pi2AES - better imaging particularly with Pepper adams Sax moving in the field. I don't know if he has actually moving, but thats what it sounds like to me. Easier to follow rhythm, likely as the bass is better defined. I think instrument layering and separation is somewhat better too.

2. Music: Spoon : Don't you evah/Rhthm and soul
much harder to tell a difference. Imaging is similar and in some areas the bass/instrumentation is better layered in the pi2AES. I prefer the pi2AES marginally as the song appeared more cohesive on the whole.



Disclaimers:
- This is not a like for like comparison since it's really a comparison of the streamer + powersupply combo. The power supply voltages are not compatible with the other streamer.
- I had to switch cables around for the comparison which takes around a minute at least. My DAC only has 1 SPDIF input.
- my speakers have an inherent limitation in center imaging where it can be a bit hazy.
- my vintage onkyo amp is ungraded with modern caps but still has the original power transistors ( is that what those things are called?)
- I almost feel it is better to do comparisons like this on unfamiliar music as its easy to get lost in music you like.

 

Do more songs! I was enjoying this comparison and I think it will be helpful for other people and is in the spirit of RD's thread.

Edit: or like you said, listen to more unfamiliar music and just do a breakdown of both streamers, however brief.

 

.

Thanks - I will do some more tomorrow

 

I think there are two things that are better:

1. Streaming options, particularly without a server. I'm comparing SqueezeServer/LMS to Roon, because they're similar, but... most people don't use Roon. Spotify Connect, Tidal Connect, or Chromecast/Airplay let random streaming services work with a lot of endpoints today, in a way that was impossible 15 years ago.

2. Control. You controlled the Squeezebox either from a web page or from a remote, looking at the VFD screen. Doing stuff on a phone/tablet is way nicer. (The Squeezebox Duet remote -- basically a small computer with a screen and scrollwheel -- was an attempt to try modernizing control schemes, but it had a lot of flaws.)

For sound quality, I think that local files have not improved, but streaming has (that Rhapsody was either AAC or MP3, not the FLAC that Tidal/Qobuz can do).

 

Ahh. My answer will show that I am a Luddite stuck in the past.

I have 4 RPi based streamers in my home, all currently running PiCoreplayer which uses Squeezelite as a player on all Pi's and LMS as a server on one of the Pi's. Although Squeezelite/LMS have evolved, they are essentially the same as they were in those early Squeezebox days. My systems are of varying quality (I don't need the best quality in my garage, for example) and I use each player differently, but I don't use any paid streaming services. Instead, I still buy CD (and LPs for my analog branch) and rip them to an SSD attached to my server RPi. I like playing random from my music or choosing a specific album to listen start to finish. I also like listening to internet radio because I find new and sometimes obscure music, even though I pay a small price in the quality of file being played. Streaming services like Spotify may have access to a very wide range of music, but I don't know how to discover new music that way, so they haven't worked for me.

Probably the biggest advance since the Sqeezebox Touch days was that you couldn't attach a hard drive to the Touch, instead I had to turn on my computer to access my music by WiFi to the Touch. That was one step I could do without.

As an aside, I have used iPeng on my phone to control Squeezelite/LMS for 10 years or so. It's not perfect, but I don't see other interfaces as better. In fact, I prefer having a dedicated app over having to go to a webpage over a browser. Fewer steps.

I keep following different approaches, both RPi and bespoke, but most emphasize paid streaming and ignore internet radio. That said, I am a hardware junkie. If I felt I could make a worthwhile gain with different hardware, I would probably try it, but I wonder if jitter improvement beyond Pi2AES is a worthwhile audible improvement. That's why I am following this thread and interested to read some members differences in sound between different streamers.

 

Only you experience what you hear. So I'll try to explain my understanding of Adapticlock based on a fair bit of research a while back.

A useful metaphor is that the streamer providing data is like an author submitting a manuscript. With a Gungnir or Yggdrasil, Adapticlock reclocking is like having a typist rework the manuscript to a standard format. Whether the handwriting is messy or beautiful, whether the i's have little hearts and the 7's have strokes doesn't matter - it is all left behind. The typist will provide output from any readable manuscript that presents the same.

Like the typist passes on the text but ditches the handwriting, Adapticlock passes on the audio data and ditches the stream's timing. For a decent source, that should improve it. Potentially, a super clean source will have its stream downgraded. Provided the stream is free of errors, you will get exactly the same output to the DAC chips.

 

A question I have about streamers in general, idk if it fits here, but thought I'd post since this thread is getting so much traction:

If a streamer—say a Neo Stream—uses femto clocks, and my DAC uses VXCO reclocking, does that mean if I plug that streamer into my DAC, will the reclocking done by my DAC internally be superseded by the streamer's clocks? Idk if this made sense, please tear this to shreds if it doesn't lol.

 

Very interesting question. If I had to hazard a guess I would say that the DAC's input buffer completely ignores the incoming clock from the streamer. Mulling it over some more... one would expect the DAC's clock would be the one primarily controlling and pulling both the source data from the streamer and the clocking out analog signal in the actual D/A process thus making the streamer's clock completely irrelevant in this particular situation. Looking forward to the experts chiming in on this topic.

 

By definition reclocking implies that the DAC is taking precedence, right?

 

The Duet was rather horrible. What's more, if I remember rightly, it cost more than the squeezebox. Fancy design, very poor actual-use experience. I mostly used the web interface on the computer.Mobile-phone apps followed, as already mentioned.

In the world of one-man-one-computer, it makes no odds. And my "well f**k off out of this thread then" attitude is I neither want nor need another computer, especially not for playing (or streaming or whatever word it is) music.The one I have has vast amounts of

Back in the early days of my computer-audio experience, it was close enough to be wired/toslinked to the rest of my hifi. It was another "Tape" device attached to the amplifier, with music going both ways. In those days, "wireless" was not such a ubiquitous term (except for old-fashioned radios), but that is what my Squeezebox was: I needed to go wire-less. Of course, there was only "O" and no "I"

Logitech Media Server, Squeezebox, etc etc: the forum seems to still be active, if anybody is interested. I haven't looked at it in ages, but I still have the bookmark.

 

A clock in digital equipment is the step timing source, i.e. it tells any kind of chip, be it a processor or RAM or whatever, when to execute the next processing step. It’s important that these steps are aligned between components. Imagine a RAM chip with a timing different from the CPU: It would put data on the connection lines (called a “bus”) to the CPU at times when the CPU doesn’t expect it to, and no communication would be possible.

If the sending and receiving side don’t use the same clock source, in the case of a self-clocking signal as we have here, the receiver must use synchronisation information embedded in the data to determine the clock and adjust its own clock accordingly (clock recovery).

No clock used in digital equipment is ever absolutely accurate, temperature makes a difference, e.g.. With the methods used in digital equipment, those differences are not a problem, the digital data can be recovered without errors unless the differences are really severe.

However, when the data is converted to analog using a DAC chip, and the clock of the DAC and the signal differ, this can lead to audible artefacts. This is also true if the clock that was used in the ADC during production differs from the clock in the DAC.

This is why clocks used in good DACs are much more precise (TCXOs or OCXOs) than those in computers. Let’s hope that those used in recording equipment are similarly good.

This much I know, the rest that follows now is speculation based on how I think it must be:

If the signal is directly fed to the DAC chip without reclocking, the timing of the clock in the DAC and the clock in the streamer must be as close as possible to avoid jitter, i. e. the more accurate both clocks are, the better the music reproduction will be. This is where a streamer can be beneficial.

If the signal is buffered first (in the DAC device) and then sent from the buffer to the DAC chip, the buffer component and the DAC chip can use the same clock source, and a better streamer shouldn’t make a difference. That’s what reclocking the signal means.

What I’m not sure about is how widespread such reclocking mechanisms are. The Cirrus WM8804 S/PDIF receiver used in my DAC, e.g., boasts it effectively removes jitter even from very bad signals. Does that mean that it reclocks? Or what else does it do apart from reducing noise, if anything?

The Zen Stream I have since some two weeks now is definitely a big improvement over the Pi4 USB I used before. Does that mean that no reclocking is happening in my DAC? Or are there other factors besides the clock that come into play here?

 

All audio inputs, other than direct I2S, use "reclocking" This is because the AES 3 stream does not have a clock. It encodes the data and the source clock into a single signal. The DAC decodes this stream and recovers the clock using a PLL (Phase Locked Loop). This works because the encoding scheme insures there are enough transitions in the data stream to continuously "reclock" to the source clock over a period of time (usually milliseconds).

Yes, I know, TMI! But, the upshot is that every DAC does this reclocking, but some do a better job than others. Also the source clock is very important since it creates the encoded stream that the PLL is locking on to. Too much jitter and the decoding circuit can not track as accurately and cause jitter in the DAC. Both sides are critical.

Where all this gets a bit trickier is if the DAC uses buffering to hold multiple, recovered samples, and then clocks them to the Digital to Analog (D/A) circuit with it's own local clock, not the recovered clock. This is what a sample rate converter (SRC) does even if the clocks are matched (ie 48Khz in, 48Khz out).

Finally, with I2S there is no recovered clock because the data and clocks (Bit Clock, Word Clock and Master Clock if used) go direct to either D/A circuit or to the SRC.

If your head has not yet exploded then you just might be positioned to make an informed decision, not one from from extreme "nervousa"!

 

My head exploded.

Can you comment on the Yggdrasil A2/og with AES specifically @Michael Kelly and @zottel ?