Hi,
The motherboards have started coming with good audio performance these days with soundcard outputs claiming 115dB SNR. I am looking for such a motherboard that has SPDIF input. Can you please provide a few model numbers if you know?
Thanks and Regards,
Goldy
The motherboards have started coming with good audio performance these days with soundcard outputs claiming 115dB SNR. I am looking for such a motherboard that has SPDIF input. Can you please provide a few model numbers if you know?
Thanks and Regards,
Goldy
I have never seen a motherboard with SPDIF input, only SPDIF output.
You could use a USB SPDIF receiver like one of these (Toslink only):
https://www.minidsp.com/products/usb-audio-interface/mchstreamer-box
https://hifimediy.com/product/s2-digi/
If you want SPDIF input and analog outputs with high sound quality just purchase a USB audio interface and forget about the motherboard audio!
You could use a USB SPDIF receiver like one of these (Toslink only):
https://www.minidsp.com/products/usb-audio-interface/mchstreamer-box
https://hifimediy.com/product/s2-digi/
If you want SPDIF input and analog outputs with high sound quality just purchase a USB audio interface and forget about the motherboard audio!
Hi Charlie,
Thanks for your quick reply. I am concerned about the latency of USB input and output. I am getting audio via spdif of my TV so lip sync is important. I have no way of delaying my video on TV and hence looking for a low latency solution.
There is a PCI sound card that can do SPDIF i/p and good quality analog out.
https://www.amazon.in/CM9882A-Bracket-Compatible-Surround-Equipment/dp/B0CF1243YJ .It claims 106dB SNR.
Do you think if I use this in a linux motherboard and run DSP-EqualizerAPO in linux, can I get better than 10ms latency for 48KHz SPDIF processing?
My linux/alsa friend tells me that with current state of high speed CPUs, its feasible by making buffers small. What do you think?
There is another way of doing it using RPi. I could use HiFiBerry Digi I/O HAT that has spdif/coax input on RPi and RPi can do audio DSP in EqualizerAPO and output can be had via HDMI to analog using https://www.amazon.in/Tobo-Digital-Converter-Support-Blue-Ray/dp/B07Y9SY7XQ. Both I2S (SPDIF) and HDMI are native in RPI processor, removing one place of latency, only DSP latency exists.
Which do you think can give me better latency? I am open to both.
Also, can you please tell whether audio dsp is done in h/w or s/w in both x86 and RPi?
Thanks and Regards,
WonderfulAudio
Thanks for your quick reply. I am concerned about the latency of USB input and output. I am getting audio via spdif of my TV so lip sync is important. I have no way of delaying my video on TV and hence looking for a low latency solution.
There is a PCI sound card that can do SPDIF i/p and good quality analog out.
https://www.amazon.in/CM9882A-Bracket-Compatible-Surround-Equipment/dp/B0CF1243YJ .It claims 106dB SNR.
Do you think if I use this in a linux motherboard and run DSP-EqualizerAPO in linux, can I get better than 10ms latency for 48KHz SPDIF processing?
My linux/alsa friend tells me that with current state of high speed CPUs, its feasible by making buffers small. What do you think?
There is another way of doing it using RPi. I could use HiFiBerry Digi I/O HAT that has spdif/coax input on RPi and RPi can do audio DSP in EqualizerAPO and output can be had via HDMI to analog using https://www.amazon.in/Tobo-Digital-Converter-Support-Blue-Ray/dp/B07Y9SY7XQ. Both I2S (SPDIF) and HDMI are native in RPI processor, removing one place of latency, only DSP latency exists.
Which do you think can give me better latency? I am open to both.
Also, can you please tell whether audio dsp is done in h/w or s/w in both x86 and RPi?
Thanks and Regards,
WonderfulAudio
Last edited:
IMO if low latency is your key criteria, the chain should be as simple as possible. Since SPDIF is clock-master input, using another clock master for output will require adaptive resampling which again introduces latency, unless done in hardware.
The standard method would be using a hardware DSP chip.
IMO a HW+software path with potential for lowest latency is SPDIF receiver I2S master > I2S slave input -> DSP on real-time linux kernel -> I2S output slaved to the SPDIF receiver I2S clocks -> I2S DAC or SPDIF transmitter. Basically similar to the hardware DSP chip, just doing the DSP in software.
Any ARM board with I2S input/output will do, e.g. RPi or countless other options from non-RPi world.
Advantage of the hardware DSP chips is that typically they have low-latency adaptive resampling which adapts automatically to incoming SPDIF->I2S samplerate. For the solution involving linux you will either be stuck at one input samplerate, or will have to communicate with the SPDIF receiver (via I2C) to read updates of incoming samplerate (most SPDIF receivers can provide this info if integrated properly).
I do not recall any project here or on audiosciencereview.com which would handle this reading SPDIF rate (eventhough it would be quite uncomplicated). To implement some linux skills would be necessary. But for the fixed rate the software part with CamillaDSP would be quite simple, a few projects like that documented here and there. All you need is SPDIF receiver I2S master and DAC/SPDIF output I2S slave. IMO this board for 15USD could do both https://vi.aliexpress.com/item/1005007975853597.html , it even has I2C control header for proper communication with the SPDIF transciever. Along with e.g. RPi Zero 2 it may work nicely.
The standard method would be using a hardware DSP chip.
IMO a HW+software path with potential for lowest latency is SPDIF receiver I2S master > I2S slave input -> DSP on real-time linux kernel -> I2S output slaved to the SPDIF receiver I2S clocks -> I2S DAC or SPDIF transmitter. Basically similar to the hardware DSP chip, just doing the DSP in software.
Any ARM board with I2S input/output will do, e.g. RPi or countless other options from non-RPi world.
Advantage of the hardware DSP chips is that typically they have low-latency adaptive resampling which adapts automatically to incoming SPDIF->I2S samplerate. For the solution involving linux you will either be stuck at one input samplerate, or will have to communicate with the SPDIF receiver (via I2C) to read updates of incoming samplerate (most SPDIF receivers can provide this info if integrated properly).
I do not recall any project here or on audiosciencereview.com which would handle this reading SPDIF rate (eventhough it would be quite uncomplicated). To implement some linux skills would be necessary. But for the fixed rate the software part with CamillaDSP would be quite simple, a few projects like that documented here and there. All you need is SPDIF receiver I2S master and DAC/SPDIF output I2S slave. IMO this board for 15USD could do both https://vi.aliexpress.com/item/1005007975853597.html , it even has I2C control header for proper communication with the SPDIF transciever. Along with e.g. RPi Zero 2 it may work nicely.
@phofman All true, but since he said "motherboard with good audio performance" I assumed he was not interested in an SBC like an ARM board but rather something more like a traditional computer with onboard sound. But in any case, you are giving some good advice now that there is a concern about latency.
Charlie, I agree my answer is not about an x86 motherboard. As you said a mb with SPDIF input is unlikely to source. While many HDA codecs do have SPDIF input features, I have not seen a mb to have these available physically as well as configured by the driver.
Then there is the issue of merging I/O clock domains (which most users do not recognize and then are disappointed). No way to avoid ASRC with HDA or USB as these are always clock masters on output (while SPDIF is always clock master in input). IMO only I2S (unavailable on x86) provides that option. Since Goldy himself talks about linux, it makes no major difference whether running linux on x86 or ARM, the methods are similar.
IIUC your gst project would fit the bill too, since no ASRC is required.
Then there is the issue of merging I/O clock domains (which most users do not recognize and then are disappointed). No way to avoid ASRC with HDA or USB as these are always clock masters on output (while SPDIF is always clock master in input). IMO only I2S (unavailable on x86) provides that option. Since Goldy himself talks about linux, it makes no major difference whether running linux on x86 or ARM, the methods are similar.
IIUC your gst project would fit the bill too, since no ASRC is required.
If you are playing video on your PC, you could use a video player that has a lip sync feature like JRiver or VLC.
Do those solutions for video and audio synchronization work for a live video stream or only when you are playing from file?
VLC is for file playback only. I am unsure about whether JRiver's lip sync feature works for a live video stream.