In our little mini-series on streaming audio, we’ve covered the overview, the server, the controller, and today let’s finish up with the final puzzle piece, the renderer.
The renderer is sometimes a separate box or card (like PS Audio’s Bridge or AirLens), or part of a more complete grouping of the major components needed to stream music. Its job is to connect with the server, accept the digital bits being streamed to it, convert those bits to a form acceptable to your DAC, and pass them along.
The renderer is the player. *(though this can be confusing because typically, a player produces something we can hear—like the output of a CD player. Renderers are digital in and digital out.)
In some parlances, the renderer is also known as the endpoint.
From a sonic standpoint, the renderer has the most important job of all. For it is here, in the final puzzle piece, where the proverbial rubber meets the road.
If we think back to our streaming system’s architecture, we remember that the server is a big network-connected hard drive. Our controller (like Roon) talks to both the server and the renderer and connects the two together when you select a track of music.
What’s important to understand is that big hard drive in the sky is sending its digital bits over the internet through a crazy combination of switches, wires, satellites, fiber, coax, etc. There are no clocks to get messed up. It’s just millions of little packets of data swarming around like bees converging in the hive. They all know where they need to go but how they get there and in what order doesn’t matter.
Our renderer organizes the swarm of bits into a uniform, orderly stream, processes the data into a form the DAC wants, then adds that all-important clock to run everything.
That final clock is where all the magic happens. Get that jitter and noise free and we have perfection.
Skimp on this last step and….
The renderer is the single most important sonic piece of the puzzle.
Make sure it’s up to your standards.
You must be logged in to post a comment.
© 2023 PS Audio, Inc.
Paul until the last few days I have never seen someone equating a server to a hard drive
That’s interesting, Willem. When I walk into PS Audio’s server room, all I hear is the noisy fans of our big network servers. They are giant many disc hard drives. And server farms are all hard drives and the computers that run them.
Funny thing is, I have never heard of them being called anything else!
I’ve known a couple of people who design them and lots of consultants who manage them. I understand server farms and data centres are quite different things. A server farm will typically host, for example, a cloud computing system (Citrix, that sort of thing), the data will be elsewhere. The hardware is different. Even in a domestic set-up they are best kept separate.
Paul please humor me for five minutes. Sign into amazon and then:
1. Search for “hard drive”. Check through the results and see if any server comes up
2. Search for “server”. Check through the results and see if any bare hard drive comes up.
Servers do contain hard drives, and to function they need storage capacity, but they are not hard drives
Paul or as an alternative to doing the amazon search you can read (in about five minutes) about the difference between a server and storage here:
Thanks, but I truly get it.
I am doing my best to keep it simple and sometimes that gets clouded up with specifics. Always a trade off, eh?
Paul i promise this is my last comment on this issue but for me calling a server a hard drive is not simplifying it, but confusing it. An exact anology would be to refer to the PS Audio power regenerators as transformers, as in i have my audio equipment plugged into one of the PS Audio transformers
It’s actually worse than Willem suggests. All the audio servers I’ve used are pure software – Linn Kinsky Server, Auralic Lightning Server and Roon Core. All they require is a piece of hardware with the appropriate operating system. It can be a laptop, a NAS, a proprietary streamer, but mostly cloud-based server systems.
Like you Willem I’ve never heard of a server equating to a hard drive.
What I see is a computer accessing the hard drives or for that matter any storage medium. That’s what I learned a server was. In a broad stroke you could say the hard drive is the server…. Walk into a computer store and say you want a hard drive to use without an enclosure. It’s not much use except as a paper weight – that is until you connect it up and then you can call it all a server.
Yes, I think not only due to usual IT terms, this needs to be separated, but especially in the topic we have here, simply, because even there, storage often is a separate entity to a server, in modern IT environments anyway (especially in case of cloud streaming service providers we speak of). I think a too rough simplification doesn’t help when as in case of our topic audio streaming, a nameable part of streaming setups use a storage separated from the server function.
(running time 3:49)
I am using a Nuc11 intel i5 where I have my Roon (ROCK) directly connected to my DAC with an excellent sound.
So in this case ROON software is acting not only as a Controller to manage music coming from Server but as well as a renderer. It works fine with an very good external power supply.
Is it a wrong stup?
No. Many single components are setup and can undertake the controller and renederer role. Sometimes rather well. My experience though is that separate components for each role if well chosen will provide a better outcome
Definitely not “wrong,” as it’s the way many people will setup.
If you don’t have a DirectStream MKII with its galvanically isolated USB, I would recommend at least getting a USB isolator like the Matrix. Not expensive, but it does help.
So the DS MK2 only galvanically isolates the USB input? That would make the USB connection the preferred easy set up to utilize the main selling attraction of the DSMK2.
I’m happy with the original DSDAC using the I2 inputs. Not as happy with it using the coax input.
Mike, It is my understanding that all of the inputs on the MK2 can be galvanically isolated.
Can be or are Tony? ✌️
The Mk2 has switchable galvanic isolation.
You can switch it ‘on’ or ‘off’…no kidding!
So Martin, what’s the benefit of switching them individually on and off?
There has to be a reason otherwise it would not be part of the design.
For comparison purposes?
In a way.
So, the way Ted explains it, it’s so that the listener can select what sounds better.
Well that is surprising because the way galvanic isolation is spoken about in these parts gives the impression that it’s the best thing since sliced bread, the bee’s knees or the db’s…..and that’s not short for decibels.
Believe me I was also surprised when
Ted uttered that piece of information 😮
I agree that good power supply (internal or external) gives extremely high quality audio.
Surely the easiest way to isolate the data from electrical power is to use wireless or fibre cable. Are they not by definition completely galvanically isolated? I would have thought a wireless streaming card counts as renderer.
I think it’s just the audiophiles’ predilection to use copper wire to send data.
Holy conspiracy Batman.
The following two articles were published by their respective associations on the 9th March. The following day Paul began his series of articles about streaming .
Coincidence or perfect timing on Paul’s part?
The given figures are for the whole market and streaming is clearly the most popular choice but I wonder how the picture might change if it were compiled only from audiophiles. My guess would be a roughly equal split between vinyl and CD with streaming in the minority, for the moment.
I still see streaming as a relatively new concept for the consumer but much less so for music moguls who have always been interested…..in the revenue stream.
I referred to the RIAA report last week. I mentioned that there were 92 million paid subscriptions in the USA in 2022. Total unit sales of vinyl and CD were 77 million, less units than the numbers of people paying and accessing thousands of tracks annually. So I reckon streaming accounts for well in excess of 99% of paid music consumption by tracks played. Far more is accessed for free.
The iTunes Store was launched in 2003 and YouTube in 2005. So mainstream streaming is about 20 years old. The CD is now 40 years old and dwindling fast. I would say streaming is totally dominant and very mature, certainly not new.
I keep reading about a resurgence in CD sales…it must be fake news.
Those reports stated a decline in CD sales in 2022 of about 20%. In the USA less than 1 CD was sold in 2022 for every 10 people.
Does this number of CD sales include purchasing them used on sites like Discogs Steven?
In my first ever job commencing in 1980 I did rendering on various structures on Tektronix machines. My son is an industrial designer and he did the rendering of my ceiling speakers, because he worked for the company that made them and he did all the product literature. If someone can find a link on the internet that describes an audio renderer, I’d like to see it.
I count 37 devices in my home that stream audio (add Alexa Echo Dot to my previous list and excluding mobile devices and computers), they all receive data wirelessly and 36 of them output sound. The only one that doesn’t outputs analogue at speaker terminals. Not a single one has a digital output. I don’t think any of them have any internal wiring, so I don’t see where I2S comes in to it.
Microsoft describe an endpoint.
This is important to know because a server is an endpoint and the most vulnerable part of any streaming system. Mine got destroyed by ransomware in 2021, fortunately my server and library are separate.
Of what we know about the AirLens, it is what is known today as a Network Media Transport. The same as a CD transport, it outputs data to a DAC.
3 days to get to the point that it all
comes down to the render… and the DAC – another piece of circuitry (computer) either built in or separate to get it to the DAC language. Kind of an anticlimactic ending.
The how to’s- best ways to’s, – the apps – the OS’s – the hardware to run the OS’s – and so on can remain the mysterious to some, and obvious to others.
I gotta go the library is getting very slow.
I’m keeping a lookout for low flying servers, but the cloud (pun intended) is very low today.
All I gain from this, and the data linked by Richtea, is that most of the world’s population that has access to the internet worked out streaming years ago, it’s only a few ageing audiophiles that are struggling.
Not struggling as much as Ms Graham interviewing Hugh Grant last night, which was priceless, a classic live TV moment. Not sure what more you could expect when a fashion model interviews a guy who read English Literature at Oxford. Reminds me of a wedding reception in New York years ago, a lady came up to my wife and her opening line was “who is you favourite designer?”, to which she replied Marks & Spencer, which is probably where Hugh Grant got his suit. You need balls to call your own face a scrotum on global TV.
And rather than struggle, some of us just don’t
care about the complexities of streaming.
Steven, I do hope that you use a very good moisturiser.
Steven, I do so love being stereotyped, especially when it’s true:
“All I gain from this, and the data linked by Richtea, is that most of the world’s population that has access to the internet worked out streaming years ago, it’s only a few ageing audiophiles that are struggling.”
BTW, I too started my first real job in 1980, however, I was a bit older than you were.
Hugh Grant had a set of “Brass Ones” when he got caught in the back seat of a Limo with a Hooker, quite a few years ago. When interviewed by Letterman within weeks of his misdemeanor arrest, Dave had to ask the obvious question… “What the hell were you thinking”. Hugh just took it in stride.
He gets respect because he’s a brilliant character actor. A Very British Scandal in 2018 was one of the best things in years.
Dave Chappell : “ I get my wrinkles out with Botox” (nudge nudge)
I’ve been led to believe that ‘Black don’t crack’.
Paul, by a short cross check in the forum I verified, that so far you nearly exclusively talked of the Airlens as a “streamer”.
You now avoid to use the word “streamer” since three days, instead you talk of “renderer” and the function coming along with it “end point”. Is there any special reason for not talking of “streamer”, as also on the hidden Airlens product description you just name it “interface”? Will you sell it as Airlens, the (network) “interface”?
It’s all part of Paul’s cunning marketing plan to confuse
us to the point where we will just give in to the confusion
& go & buy a f#@king ‘Airlense’ just to shut him up.
(love you Paul 😉 )
It’s the most descriptive way to reach the idea to as many people as possible.
In our posts and discussions, we are getting to the nitty gritty of servers and streamers and renderers, oh my! But these posts appeal to a smaller more technically curious group of people.
For the majority of our customers, they are more like what Steven describes. Simple. Keep it simple.
You want to add streaming to your DSMKII? Add the AirLens streamer.
Now you can stream.
Imagine if I had called it an AirLens renderer. 🙂
I’d have more questions than answers.
It’s like calling the accelerator pedal on my Tesla a gas pedal.
That’s a lot more expressive (the totally incorrect) than anything else I could say.
My goal is to simply communicate.
Thanks yes…the simplifying approach was why I wondered you called it renderer instead of streamer the last days…I thought most of those who like it simple can do more with the term “streamer”, as I never saw anyone talking of renderer except maybe when going to the nitty gritty..that’s why I looked for the connection of the nitty gritty term to the simple term all others use.
So when we try to get a common understanding here, clarifying renderer=streamer isn’t wasted time I’d say.
My reality as a consumer turns all this on its head – not to ask what a device does – but list the things I want it to do and shortlist products.
Every streamer must at a minimum:
– Have network connectivity
– Transport the data with an app to control it
– Output a digital or analog music stream
Additional features are typically, for example:
– Rip CDs
– Store files (ripped or downloaded)
– Have favourite music service transports onboard (Spotify, Tidal etc.)
– Be Roon Ready
– Include a DAC for analogue output
– Include volume control (analog or digital)
– Have server software for multi-device output
– DSP and equalisation
– Compatible formats e.g. DSD
– Have internal linear power
– Have DC-in for optional external linear power
– Have optical input for galvanic isolation
– Have wifi in/out/both for galvanic isolation
– enabled with uPnP, Airplay, Chromecast: industry standard wireless protocols
– Have the required outputs compatible with the endpoint player
I’ve done this shopping list approach and the shortlist rapidly emerges.
To be honest, I thought this discussion was to be about the things that might be on people’s shopping lists, like reclocking, upscaling, etc., because specific tasks or functions tend to be the selling points of various streamer brands.
I thought the discussion could possibly have been to finally (besides the technical terms) enable folks to conclude which options they have to build a streaming setup, by clarifying how a pure streamer identifies against a pure server and a storage device (included in it or not) and the various combinations inbetween, which also includes to know what only of all this is needed when streaming from an online service.
Questions in the forum like ”I have an Innuos Zen server, do I need the Airlens“ occasionally demonstrated how lost they are even when they have everything they need.
But I got that the series was to technically explain the basic streaming process and terms, without a focus on exact differentiations as well as the setup options and role of various gear specificity, my bad, it was fine in its focus I think.
I would agree, except discussing streaming forgetting the actual streaming software is a bi5 like a horse race without any horses.
My assumed realization of yesterday, that we mainly talk about SW, was, after the day before yesterday, also today again nullified 😉 when I read that besides the streamer=HW (I agree, as the SW belongs to the server function), also the server is more or less understood as a combination of server HW with storage and library/management SW (which without doubt it can be in one or two kinds of configurations). I understood this was done to simplify (which it did if staying at the surface is ok).
The accelerator pedal on my motorcycle is the right-hand grip on the handlebar. 😎
In my system, the (digital) rubber meets the (analog) road at the output stage of the DAC. That’s where the tubes are. 😎
My new book. Very short. Titled “Streaming for Dummies and Ageing Audiophiles”
Buy streamer from someone like Nad, Cambridge Audio, Auralic, Innuos.
Place in Audio rack.
Connect to your Dac and LAN
Load app on your Android or iPhone or Tablet or iPad.
Follow the 5 or 6 steps
Select your favorite album
Repeat with different albums to your hearts content
Appendix: For better audio quality plug the power cord into your PS Audio transformer
Add to the list: Fix the streamer IP address, and when you change internet service provider and they install a new modem, remember to re-fix it.
(As you might image, last night I was wondering where all my music had gone.)
Been there, done that Steven. A living nightmare for two weeks until I got everything straightened out. FYI… The IP address was self correcting and did not solve the underlying problem of why I was unable to stream after trying to swap out my router. As it turned out it was a simple fix for Stupid Stimpy2.
Getting devices with some computing carry-on inside them to talk to each other, for just doing completely non-audio stuff, can very easily degenerate into an exercise in pure frustration. Throw in Windows, and your nightmare is complete 🙂 . I’ve been in computers for over 40 years, but still have a home network that *always* misbehaves, doesn’t work in some way, when trying to do something incredibly simple, like transferring a file.
A memory stick, and your feet, come to the rescue – I’ve given up on doing it the “right way!”, once it starts to annoy me …
Paul- you are wrong about how data is processed on the receiving server, the app/renderer has nothing to do with organizing data packets, it’s all done by the network equipment. TCP/IP is the protocol that is used to make sure that the data packets sent by the host are put together in the correct order on the receiving host. Data packets sent by (let’s say) Tidal, might take different network paths to get to your home computer. The network protocol will take these packets and put them in the correct order, and fix any corruption (issue a resend packet) before the receiving host can use the data.
Here is an explanation of the protocol and it’s function:
TCP is a connection-oriented protocol and is designed to allow two hosts to establish a reliable connection and exchange data. Contrary to UDP, TCP guarantees that packets will be delivered and processed in the same order in which they were sent.
You are correct, thanks. Better for me to have said the renderer takes the network data and converts it into a form useable by the DAC which is either a S/PDIF or I2S format.
I do have a question about jitter. Let me be clear, I know absolutely nothing about the inner workings of stereo equipment. Here goes: If jitter is so critical, why isn’t there a clocking station embedded as the very first stage in the DAC. The Mk I had the option of a network bridge. Will that option ever happen again?