Why does Volumio downsample to 48kHz on spdif ??

Hi There,

My first attempt to play Volumio on my new Cubox-12eX is disappointing.

Spdif output to NAD M51 DAC and Volumio is downsampling my 96kHz flac files to 48kHz.

Does anybody know why Volumio is downsampling and how I can prevent downsampling?


  • Tony

If you see downsampling on S\PDIF this happens most likely because the S\PDIF driver accepts no more than 48 khrtz. We could look into the alsa conf file to try to avoid it. I will have a look and report back

Hi all

I’m new here and I encounter the same issue as Scarletlion: It looks like Volumio is downsampling high resolution signals to 48kHz.
I’m using a Cubox I4-Pro which is running Volumio v1.4 with an updated kernel to have WIFI working. It works great, thanks to the efforts of Michelangelo and Esseki, thanks guys!
My Cubox is connected by means of SPDIF to a Marantz SA-KI Pearl Lite SACD player which has inputs for digital signals, like SPDIF, USB and Coax. I only tested SPDIF.

I am very positive about Volumio! It looks to me as a very promising project. I’m using it now for about 2 weeks and it works out great for me.



This is not a volumio issue, the optical spdif on the cubox-i is clocked by the main processor and limited to 24-48. Solid-run have reserved space on one of the boards for a high res external (to the processor) clock, but are not shipping with that, yet.The HDMI output is 16-96, I think.

However this leads to an important issue… Volumio and Cubox-i can play high resolution 24/192 files heck, the android tablet I’m using can do that too. but outputing that resolution as a bit perfect stream is another matter entirely. A lot of much more expensive computers and their sound cards and audio software simply don’t have the ability to do that either!

What the Cubox-i and Volumio do spectacularly well is pass 24-48 or lower signals to your DAC, through spdif. It’s amazing how awesome they sound even over wifi… (Thank you for the fix Michelangelo)

So… Assume you have a Mac mini, or similar and you can output hi res files bit perfect, you then need to pass the signal to a hi-fi dac that is capable of successfully converting the signal to analog and then pass that to a pre-amp where you will hear the difference.

Now also assume you own a Terabyte+ NAS so you can fit more than one album that ranges between 1 and 3 GB in size on and you have a wired network to your computer cos hi res audio doesn’t work over WiFi, then…

The big one…

Assume that the person who rendered the hi res file did so from even higher resolution master source, this is usually the studio master tape, or DAT, or original analog tape. And the equipment used by the musicians/studio staff was up to the job…

If you assume all those things then you can justifiably start to question the hi-res vs 16/44 - 24/48 quality issue. It’s one I’m wrestling with myself at the moment.


i’m the happy owner of a Cubox-i4Pro that runs Volumio 1.4. The cubox is connected to my DAC via USB. The problem is that my DAC does not show the resolution of the files played.

Could somebody confirm if the Cubox down sample when using a USB connection to a DAC or does it happen only with SPDIF ?

Thank you in advance

Hi I’ve same issue,

my Cambridge Audio AV receiver cannot display income quality of PCM over optical SPDIF.
Right now I’m using Apple TV, but ATV downscale audio to 48kHz ; This is reason why,
I plan to buy R-PI with optical out. Question is: will R-PI device supports true 192/24 PCM over optical ?
(sure with R-PI - HifiBerry with digital SPDIF)





Here are the threads in the cubox-i forums relating audio out via spdif, USB and HDMI.


Hi Esseki,

I have found the time to test the USB playback of the Cubox-i4Pro to my Marantz SA-KI Pearl Lite which I use as DAC. The DAC is recognized by Volumio, I am able to choose Audio Output “Audio” at the MPD Configuration Settings. This option was previously unavailable since I did not use USB to connect (I previously only tested the optical SPDIF connection).
I tested playback of all kind of music files in FLAC format. They are all playing clean as far as I have tested for now. The only thing is that my DAC shows 96KHz as the highest sampling frequency. So if I play 44.1KHz audio files then my display shows 44.1KHz, if I play 96kHz audio files the display shows 96kHz and if I play 192kHz audio files then my display keeps showing 96kHz. So it looks like downsampling on USB too. I don’t know if the Cubox or Volumio is the cause for it (keeping in mind the post of ianmr65).



Hi HM_Dutch,

Thank you for your response :slight_smile:
I’ m curious now to get the reason why the Cubox is down sampling when using the USB output ?



Envoyé de mon iPad en utilisant Tapatalk

Hi HM_Dutch & esseki,

If you look deep into the Marantz manual the reason is revealed:-

USB Audio Class 1 will only go as far as 96kHz so while your DAC is capable of 24/192 kHz; when you are using the USB Interface the maximum sampling rate will be 96kHz. It is therefore the correct function of the Volumio software to downsample 192 >> 96.

See this link for a bit more background on USB Audio Class Standards

amimuse.com/2012/09/asynchronous … class-2-0/

Hi NWT Stuff,

Thanks for your reply. I have searched the SA-KI Pearl Lite manual and you are right. The SACD players sampling frequency on USB type B port is limited to 96 kHz / 24 bit. That is a pity. The DAC however is capable of handling sampling frequencies of 192 kHz / 24 bit but not on the USB port. For the Optical and Coaxial digital in ports I can find no limitations so I assume they are not limited. However I found out, like scarletlion’s first post, that optical limits to 48 kHz but that has maybe something to do with the Cubox-i itself, see post of ianmr65.

So…as I have no means to try the Marantz’s coaxial digital input I will be limited to 96 kHz / 24 bit. Not bad. I have to admit that my first idea was that I must be able to play 192 kHz / 24 bit flac files with the cubox-i / volumio / SA-KI Pearl Lite combination but it seems this is not the case.

Thanks again.


This is the same trouble as multichannel audio downsampled to stereo.

All you need is S/PDIF passthrough. Sound should be put to S/PDIF without any resampling. This will fix both things.

MPD can do this. All it needs is proper setting.

So it might be technically possible for the Cubox-i to output hi res spdif passthrough using geebox but certainly not running volumio 1.4. The Cubox-i 4 pro just doesn’t have the hardware, oscillators, or async USB capabilties.*
The solution is to attach a hi-res USB to spdif converter to take care of the signal handling. You can pick a really good xmos x20 one up from ebay for around $60 US or similar here.

I can confirm that this works flawlessly out of the box with volumio on the cubox-i4 pro connected via USB. Outputting up to 24/192 and radically improving the sound quality of any file from 320 mp3 up. It also removes clicks and pops when switching formats and runs off the cubox usb power, so you don’t need a seperate power adapter. Though if you want to use it on a windows computer for testing you need drivers.

I’m afraid I don’t know what the jitter levels are, but the DAC with volumio and the user control panel with foobar on a laptop show the correct rates for the file type and makes volumio sounds amazing on my system!

  • Actually most computers don’t have this capability either. Even the mac mini only outputs at 24/96 over spdif. Most laptops are restricted to 16/46. The desktops that do output high res have to have speciality soundcards or a much more expensive USB to spdif adapter like the hiface2.

I have made some investigation on this. What I found is mpd uses ffmpg for DTS output and it seems to be ffmpg what does downsampling.

So - one solution might be changing mpd so, that it does not use ffmpg for DTS mimetype.

I have this trouble with Cubietruck and I’m not going to invest on some USB to S/PDIF dac since there is already one.

Well after a month setting it all up can say unquestionably that 24/192 recordings played through cubox-i4 pro \ volumio\ x20 converter \ rega 24/192 dac sound far better than any other digital source I have ever heard. Everything is sharper, crisper more detailed, more tonally colourful, it’s like the musicians are in the room with you. The soundstage is wider, the instruments are all much more distinguisable from each other… Wow!

If we can just skip the discussion of whether one is able to hear high res benefits, I can state from experience that Cubox-i is definitely capable of outputting 192kHz @ 24bits over Toslink.
Latest Arch has (recent) patches to resolve the lower res issue. So michelangelo’s assertion was coorrect.

I think that whilst people have managed to output hi res over toslink from the cubox-i using the geebox distro for example, the output is severely hampered by the absence of hardware clocks and ossilators

Here are some results from such native output.

32000Hz -> 32005Hz, error 0.016%
44100Hz -> 44132Hz, error 0.073%
48000Hz -> 48008Hz, error 0.017%
96000Hz -> 96017Hz, errorr 0.018%
88200Hz -> 87719Hz, error -0.544%
176400Hz -> 177631Hz, error 0.698%
192000Hz -> 192034Hz, error 0.018%

Ideally you need three clocks one for 32mhz, one for 44.1mhz family, and one for the 48mhz family. Cubox-i has no such clocks so outputting is software driven and subject to error and poorer sound quality as a result.

Well, the topic was the downsampling and its cause, merely wanted to set the record straight.
Re the clocking: those figures -but I get different ones- had already been posted July 18 in the Solidrun forum and I do think it was a big mistake to forego the ‘silab5351a’ clock generator that the blueprints show and instead rely on the i.MX6 internal clock. I for one was disappointed when I found out as it is a step back from the previous Cubox design.

It does seem possible to correct those deviations (if I understand Triodes remarks on the subject correctly -he refers to similar issues with the wandboard BTW) and whether they’re detrimental to sound quality subjectively is another matter altogether.

I don’t normally throw ‘science’ around, much prefering people to trust their ears but in this case it’s Important.

If a signal has been accurately recorded at 16/44.10 but outputs at 16/44.13 you have sampling jitter which does effect sound quality very audibley… Audio systems are prone to jitter of different kinds in different parts of the chain. This is why people spend lots of time and effort in reducing jitter through transmission (cables and connectors) and sampling (accurate clocking) and so on.
What you are trying to achieve is that exactly the same wave that exsits in the recorded material ends up, albeit amplified, driving your speakers. That is the definition of HiFi ie as accurate as possible reproduction of recorded waves to sound waves. So if cubox-i volumio is outputting inaccurately clocked signals something has to put them right. Either an asyncrnous USB DAC. Or as in my case an asyncrenous USB to spdif converter with it’s own clocks, and an asyncrenous spdif DAC with it’s own clocks…

This is a problem across all computer driven HiFi not just the cubox. The computer does an amazing job of outputting bit perfect streams, much better than CD players can, because the stream is not dependant on cd disk speed or anything else mechanical. However clocking the accuracy of the timing of the delivered bits, it’s very poor at. Unless it has specialised audiophille soundcards and audio circuits, or external clocking.

Here is a great article on jitter the ‘digital enemy’

Not quite sure what you’re reacting to? Never said anything about Cubox-i being correct in its output, far from it, as I had read Rudi’s post I’m referring to before -in July to be exact. And of course such inaccuracy is very much unwanted in any HiFi rendering device, that’s rather obvious.

Whether a defect is audible totally depends on the listeners hearing abilities. I will not speculate on those.
If we google ‘blind test jitter’ we get a host of results indicating it’s not so bad as it’s always made out to be. I especially recommend the piece by Nwavguy. :wink:

Yes, of course reclocking hardware can resolve the issue -as was claimed in the second link I posted- but I don’t think I’m the only one who thinks cluttering up the sheer elegance a single tiny box setup by hooking other stuff up to it is an unwanted solution.
Apparently Triode thinks reclocking can be done inside the Cubox through software means, I’m really interested in the way this would work out.
The possibility itself would suggest the clock itself is accurate enough, but rather it’s the dividing/multiplication algorithms that aren’t up to par :question: