Comment by 🚀 stack
Re: "Anyone working on any non-CPU projects?"
That sounds interesting. I don't understand how you plan to denoise data off a cassette in the first place -- it will be full of glorious noise along with wow and flutter from your reading deck...
But come to think of it everything I've ever done with FPGAs is cpu-related, including a full Apple ][ clone and a bunch of 6502, 6809, and custom stack machines
Jan 08 · 4 months ago
6 Later Comments ↓
🤠 shaved_yak [OP] · Jan 08 at 02:46:
Yeah, there are lots of "if"s in the physical format. My plan is to overdrive it while recording onto the tape, so hopefully it'll still be above the noise even with a 20kHz bit clock. Only the zero crossings would matter, so Schmitt triggering might help. Wow is around 1Hz and flutter maybe 1kHz, so hopefully timing can be recovered well enough. 8b10b encoding would keep it zero balanced and yield 16kbps from a 20kHz bit clock, so two stereo tracks would yield 32kbps, which is the lowest rate in the mpeg1 standard. If 20kHz ends up being unreasonable, some kind of qam might do the trick. But for now, I'm working on the tape effect pipeline, which is about done.
🤠 shaved_yak [OP] · Jan 08 at 02:49:
My own cpu was a stack machine just to get something simple running quickly, but I fell in love with the architecture. Stack machines are some of my favorites now. Have you seen the GreenArrays chips? A grid of computers on an fpga would be a cool project too.
For your project, it has happened before as Philips DCC
— https://en.wikipedia.org/wiki/Digital_Compact_Cassette
DCC used 9 tracks to store the data at 384 kbit/s so my two cents would be to store the music in he-acc and at 64bit/s on two tracks but that is just a guess.
🚀 stack [mod] · Jan 08 at 20:32:
I still don't understand how you plan to de-noise the original music. The original signal on the tape is noisy, and no amount of equipment or high sample rates can help -- the best you can do is capture the noise and not add too much more. But you will get a lot of noise and a pretty weird frequency response from the tape.
Or am I totally missing what you are saying and you are talking about storing music in a weird format on tape?
I guess I am totally confused by what you mean by reading music from cassette as digital data. Is it stored as digital data by you in the first place?
🤠 shaved_yak [OP] · Jan 08 at 20:48:
That's a neat format! I hadn't heard of it before. It doesn't really fit with my project though, since I'm going for more of a joke rather than anything practical. I thought the idea of storing a modern digital music format on an old analog one then intentionally ruining it with effects that make it sound like a tape again would be funny. Kind of like a musical successor to Tom7's reverse-emulated NES.
I'll be keeping an eye out for DCC equipment though... That would be cool to play with.
🤠 shaved_yak [OP] · Jan 08 at 20:50:
stack: oh, maybe I didn't explain it well. I'll be starting with a normal, high quality audio file and not original tape music. I'll be storing it digitally on the tape, and the digital data can be denoised because of the quantization.
Original Post
Anyone working on any non-CPU projects? — Most hobbyist FPGA projects that I've seen are homebrew CPUs or emulators for old computers or game platforms. They're cool projects and a rite of passage (I've got my own too), but I'm curious if anyone is working on anything totally different. My current project is an attempt to take audio off a cassette tape, decode it as digital data, then treat that as a stream of mp3 data and decode it on the fly - playing music without the noise, high-freq...