Great work, looking good.

Another quite new discovery for me was that apparently ESP32 is 5V tolerant. And actually LPT doesn't give 5V signals but little under 5V.
So I soldered dupont cables to LPT-connector and connected them directly to ESP32 pinheader. For straighforward audio output I2S DAC is easiest.

some later laptops output only around 3,3V on LPT.
This was giving me quiete some Headache figuring out why my device didnt work.

Cool project! Finally an open-source parallel port device that supports both Covox and DSS modes.

I was wondering whether something like that could be done with an RP2040 microcontroller. It already has two CPU cores, so the same bitbanging trick could work with that architecture as well, right? And perhaps it has the needed reaction speeds, so you wouldn't have to use such a workaround on that at all?

Here's hoping that Dreamblaster develops and starts selling a "CVX5" with DSS compatibility, based on your design, or something similar!

Thanks for sharing. 🙂

I have rewritten code and started over countless times and maybe this time this is last time. I ditched Arduino IDE in favor of getting all interrupts and watchdogs totally off. Other core runs now only loop with pure assembly without any interruptions. This was needed to get StereoIn1 (Crystal Dreams, Inertia Player) to work. Signal for channelselection is ridiculous (scope image in attachment) and sample read must be very fast. I didn't get it to work until all my software tweaks and added quite strong pullup resistor for channelselection-pin. It is much smaller and shorter than DSS-signal, so now DSS works even more realiable, as it works too without any interruptions in very tight loop. Now Crystal Dreams works 44.1kHz at Stereo (at least with my Compaq Contura 420C 75MHz) and it sounds stunning.

This detects Covox/DSS/StereoIn1-mode automatically, but StereoIn1 some software (Crystal Dreams and Inertia Player at least) puts DSS pin to HIGH, so it must be reset somehow to get back to normal Covox-mode. It cannot know if DSS signal is HIGH because DSS mode is needed or if it just because something else raised DSS-signal. I cannot force it to normal Covox mode after StereoIn1-mode, because then it wouldn't work if DSS-stuff is started after that.

https://github.com/mcgurk/Covox

I've just managed to build this thingy, without any serious skill. Only addition, that I had to set the H1L-H4L solder bridges on the pcm5102, otherwise sound was totally distorted.
Now it works like charm and it adds some live to my soundless Contura Aero.
Thanks for publishing your project!
Theoretically an opl2 emulation is possible with this design? (eg using emu8950)

It might be possible to add output support for the Disney Sound Source (and clones such as this one) to SBEMU. Not sure how well the performance would be on older CPUs, though. How efficient is emu8950? I know that for instanced Nuked-OPL3 has pretty demanding CPU requirements as a price for its "perfect" emulation accuracy.

Don't know about the exact CPU req. but a highly optimized version runs on RP2040 as part of doom port.
rp2040 OPL2 emulator (for DOOM)

and used by picogus found here on VOGONS