a 2k program, 128 bytes of ram... and 4 megabits of flash! still super cool
i really appreciate the filesystem design notes
i've been wanting to do something like this without the 4 megabits of flash; wondering if i can use lpc-10 on an atmega328, since it has a hardware multiplier (unlike the attiny45 and attiny2313s i have)
i've been thinking that it would be nice to get a speaker driver circuit that eliminates the dc bias from the speaker current that this transistor-switch driver is providing. you can't just put an ac-coupling capacitor in series with the speaker like you normally would because the cap charges up to 3 volts and stays there, with no way to ever discharge. you could of course ground the emitter through an inductor or resistor, but is there a better way?
yeah, the lpc-10 voice codec i mentioned uses the same linear predictive coding approach the chip in the speak & spell used; it requires about 10 multiplications per output sample, which is about 80000 per second, which may be challenging on an attiny2313 or attiny45 where you have to do the multiplications with repeated shifts and adds, but i think ought to be doable on the atmega328
I should have mentioned that the project is 11 years old (2012) so things were a but different (rpi were just out, no Chinese padauk-like 5cts microcontrollers but also attinys were less expensive pre covid !) And yes it uses a 512k flash :)
I'm a cheapo, I'd use a ch32v003 an stm8 or a modern attiny ;) or maybe a stm32 with 128k of flash and a dac and would compress the audio.. much more choices now ! Speech synthesis on a cheap mcu would be awesome
only about 4$ of parts [cost] have been used to build a complete amplified audio game with 1 minute Audio storage and many sentences.
For truly low-cost MCUs with similar capabilities and applications, look at Generalplus/Sunplus ICs -- sub-$1 prices, bare die/COB, and widely used in commercial products. E.g. GPC22C170A has 256B of RAM and 512KB of ROM.
2KiB ROM and 128B RAM happens to be exactly what the Atari 2600 launch titles used. Interesting that the same numbers would pop up in newer hardware projects.
Wondering if QOA would work for the audio? https://qoaformat.org/ it's a pretty small implementation with decent compression, but not sure if it'd work in those constraints
(I'm guessing not. QOA's format includes 128 bytes of lms_state per channel even before it comes to the samples; way too big)
I see an `iframe` with this url [1] but the `<body>` of the document is empty. It's very likely you have to change the protocol to `https` (instead of `http`), otherwise the browser refuses to load the url automatically.
it is that - if you turn on the setting "Enable HTTPS mode in all windows" in firefox, the http is automatically converted to an https link and it works.
i really appreciate the filesystem design notes
i've been wanting to do something like this without the 4 megabits of flash; wondering if i can use lpc-10 on an atmega328, since it has a hardware multiplier (unlike the attiny45 and attiny2313s i have)
i've been thinking that it would be nice to get a speaker driver circuit that eliminates the dc bias from the speaker current that this transistor-switch driver is providing. you can't just put an ac-coupling capacitor in series with the speaker like you normally would because the cap charges up to 3 volts and stays there, with no way to ever discharge. you could of course ground the emitter through an inductor or resistor, but is there a better way?