Options for Digital Audio Extraction and CamillaDSP

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I've been thinking through a design for my next two channel stereo system with active digital crossovers using CamillaDSP. For the player / DSP I'd use a Raspberry Pi 4b and Moode audio output to loopback, then use Camilla for mixing / crossover. USB3 connects the Pi to a Motu 8a for DAC output and ADC
input duties. The beauty of the Motu is that you can script it's routing / switching / mixing via its API over ethernet -- creating sort of a hybrid active crossover / receiver / player.

Now my question: As always the WAF factor is demanding things, so all that money spent on the lxmini+2 / Motu / tri-amp will be used when it's movie time.

So I have to transport TV / Cinema audio over to Camilla in digital format preferably unmolested by the analog and back again journey. My initial investigations show a rather arduous path from HDMI / eARC to SPDIF extractor to re-packetization using an inexpensive outboard USB sound card where it comes out as 2 channel USB audio and as a mix input into Camilla. So a couple thoughts:

1. Has anyone attempted this?
2. Expected latency issues? I know eARC has compensation for some of this.
3. Dependability of eARC because of HDCP copy protection etc. Technically since I would be getting at least a 2 channel pure digital copy I'm thinking someone would not want me to do this and it would break somewhere.
4. I also wonder whether it will work okay for some things with HT outputting 2 channel PCM over ARC, but with high endish Digital Dolby + / Atmos / etc ARC audio gets blocked during extraction because of HDCP evil.
5. If HT is processing 5.1 or whatever, will the extractor get a 2 channel PCM copy of just the LR channels?

If it wouldn't work I'd have to just suck it up and do the DAC -> ADC path through the Motu and back into the Pi/Camilla. It's only theater audio after all ;)

The cool thing about digital crossovers is how powerful they can be. The downside is that everything needs to come in digitally including analog via ADC. Something tells me that the minimal noise in such conversions would be outweighed by the large increase in performance for digital crossovers. I'd actually be more concerned about phase errors accumulating and losing holography.

Phewe. FWIW
 
If you don't need multichannel and have optical output on your TV you will probably save yourself a lot of headache using that. And the TV will handle the multichannel mixing and conversion to PCM for you. Then just look for an optical input RPI hat to avoid all USB and get the data straight to the DSP.
You will get a delay from Camilla so you'll need your TV to introduce video delay to counter.

A few years back I switched to Kodi and tvheadend for all TV/cinema duties and turned my smart TV into a dumb monitor in order to move everything to the digital domain and avoid audio having to ever go to the TV or pass over HDMI :)
 
A couple points about a hat solution like the hifiberry digi+i/o:

1. It's a hat, so the case now becomes an issue
2. It probably uses gpio, while the USB input on the rPi 4 goes straight to pcie
3. It's $50, while an inexpensive outboard USB sound card is about $25.

I think the core question in this is when you set the digital audio out on the TV to PCM does it generate L/R PCM even when its playing dolby digital 5.1? Does it mean that all other HDMI audio out will be PCM which prevents you from having access to the encoded dolby stream for your receiver.

The more I think about eARC, the more that I think that an after-market HDMI audio extractor is the only way to go. I think some of those will do the extraction and pass through the encoded stream to your receiver.
 
2. It probably uses gpio, while the USB input on the rPi 4 goes straight to pcie

It's using I2S (WM8804). RPi (the Broadcom SoC) has an internal I2S peripheral connected directly to its internal AMBA bus - see Raspberry Pi 4 Model B: Blockschaltbild des Broadcom BCM2711 | c't Magazin . Lower latency than (AMBA -> ) PCI-e -> USB controller -> USB device -> I2S -> SPDIF transciever. To process audio along with video (i.e. fixed latency required) I would avoid the USB layer.

3. It's $50, while an inexpensive outboard USB sound card is about $25.

Actually a USB soundcard with 192/24 SPDIF in/out, full linux support, and fully known internal clock arrangement (SPDIF in timing the incoming stream via USB) may be way more expensive, if available at all.

If using HDMI, I am afraid you will hit the HDCP issues (as your device is exactly what HDCP was designed to prevent). If you can get your HD digital signal to 192/24 SPDIF somehow, there are libraries which convert the HD digital formats to PCM which you can further process by Camilla DSP. Again - latency will be large and you will need means for delaying your video stream to maintain lipsync.
 
Last edited:
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.