[request] import sound tool/cmd (wave data not tracker)

[atesin]

hi... the documentation is very swallow... but i was reading the "import" console command, not too much explanatory

as far as i understood, we can open tic-80 folder ("folder" command) and once opened we can place any file we want, for each tic80 "editor":

• code = txt file?
• sprites = png/bmp files?
• map = no idea which filetype, xls?
• tracker = midi files?, mod/xm?

but what about SOUND? ... is very tedious having to plot waves dot by dot, moreover i cant imagine how it will end sounding, is more like a trial-error process... it is also very hard to try to match some sound with other "real" one, would be great if we could just sample, lower the quality and just import

would be great if import could also work with sound files (wav?), and the documentation specifies the process and wave format (frequency, bit resolution, which voice/channel, etc)

.... in case it doesnt support.. if does could you please update documentation in details?... thanks

[joshgoebel]

You don't really have much storage for something like a WAV file...

https://github.com/nesbox/TIC-80/wiki/RAM#sound-registers

You can update the Sound Registers every tick (60 times per second) to generate sound independently from the built-in API functions.

So with custom programming you could have 1.92khz (60*32), 4 bit audio... at the cost of like 1kb of storage per second... perhaps enough for like a few very specific and distorted sound effects...? Or if you gave up MAP you could store 32 seconds of audio or so... maybe like if you had an audio "cut scene" paired with a graphic? But you'll burn thru that storage fast...

[atesin]

i dont know tic80 internals, but i doubt it stores raw wave data in memory, like you i think it would consume a lot of memory.... but i think instead, it stores some points needed to "draw" the wave, and those are the dots you see when you "draw" the wave in sfx editor and tic80 read and interprets when sounds are played

i would like to know if could be have ANY way of import some sound inside tic80, not only by store wav data directly but also could be a reader/analyzer that parses a .wav file and tell tic80 what points to plot in sfx editor...

a sound copy/paste function would be terrific, or a wav file importer/analyzer command or external tool would be great also... it would greatly improve que quality of our sfxs/songs

[joshgoebel]

but i doubt it stores raw wave data in memory,

Well, it doesn't really store "raw wav data" at all - or rather it has no where specific to store such data... you could encode it into your program source, or store it in your cartridge... but to play it you have to use the facilities that TIC-80 gives you sound wise - which [most directly] are the sound registers... 32 samples of 4-bit audio that you'd have to make last 1/60 of a second... you'd have to constantly fill each frame from whereever you had the WAV data stored.

i would like to know if could be have ANY way of import some sound inside tic80,

Sure. Just downsample to 1.92kb 4-bit audio, store it in the cartridge RAM, and then write a program to read it and shove it into the sound register every frame. It would be a pretty cool demo actually.

My worry would be timing issues, that TIC-80 isn't really designed/tested to work this way and you might do all that work just to find out you have clicks or dropped audio or other strange issues... but maybe it "just works"...

[joshgoebel]

I think something like https://sox.sourceforge.net/sox.html could get you close to the raw data you'd need... output a mono 4-bit 1.92khz raw file... the you'd just need a small script to convert that raw file into textual binary data (the same way TIC-80 saves such data in text mode)... then you just need to write the audio processing code. :-)

[joshgoebel]

sox input.mp3 -b8 -c1 -r1920 -eunsigned-integer out.raw

Doesn't seem to want to write 4-bit so you'd have to also convert the 8-bit samples to 4-bit, easy enough. Doesn't sound great already though, you'd probably want to specially process the audio for such tight encoding, not just convert it "as is".

Ok, sox is amazing and if you knew what you needed it could probably do any type of post-processing you needed to try and clean the audio up...

[joshgoebel]

a wav file importer/analyzer command

This is unlikely since there is really no way to use that data without a lot of custom programming. Importing a MIDI would honestly make way more sense, etc... https://github.com/wojciech-graj/TIC-MIDI

[atesin]

i agree with you about midi... but midi are just song patterns without any actual sound wave

for example to make music, i have no idea how to create bassdrum kick or claps or cymbal sounds, or how "looks" a violin or trumpet or any other "real world" sound... i can't imagine how sounds are "seen" graphically to replicate it :(

i would like if we could sample some sounds in some ecternal editor tool like audacity, audition, soundforge, etc... process it, stretch to required freq/lenght, downsampling, export to wav.... then run a tool to read wav files, parses and put data in a .cart directly

that wav file parser+importer would be a very powerful and useful tool, we could just use all the power of some external sound editor, we could put very rich sounds inside tic80

[bztsrc]

hi... the documentation is very swallow...

I agree!

Well, it doesn't really store "raw wav data" at all

I'm fairly certain that it does, at offset 0xFFE4 (see wiki). It's just not the entire song is stores as wave, only the instruments (which then is put together into a whole song using the tracker).

but midi are just song patterns without any actual sound wave

That's for the tracker. The tracker contains song patterns, not waves either.

for example to make music, i have no idea how to create bassdrum kick or claps or cymbal sounds, or how "looks" a violin or trumpet or any other "real world" sound... i can't imagine how sounds are "seen" graphically to replicate it :(

If you're familiar with old-school trackers (FT2, OpenMPT, MilkyTracker etc.), then those work similarly like TIC-80: they have a few instruments, stored as waves, and MIDI-like patterns with musical notes. You can load a .mod file with a trumpet or violin for example and see the waveform in those trackers. However .mod files have 8-bit or 16-bit samples, and one instrument might have up to 65536 samples. On TIC-80, one sample is stored on 4-bit, and you have only 32 samples, but despite the limited precision the wave must look the same to give the same sound.

Another approach try using Audacity. Load a violin wave for example which is 1 second long, then on the bottom left enter "32 Hz" (this only works if the length of the sample is 1 sec). The resulting waveform should then look like exactly like the one you have to set in TIC-80 for a similar sound.

I'd like to point out that violin has probably one of the most complex waveform with lots of harmonics. It is highly unlikely that you'll be able to reproduce it using chip-tunes. But at least, using the methods above, you'll see how the sound's wave should look like.

Hope this helps,
bzt

[bztsrc]

run a tool to read wav files, parses and put data in a .cart directly

Ah, about that, it won't work. You might be able to use wav files for the waveforms only, and not for the entire song as stored in the .cart.

(Maybe if the song is simple, then you can locate the individual instrument wave in that, and using sophisticated 1-d pattern matching algorithms or maybe AI you can convert that into a list of musical notes required for the tracker. But it is highly unlikely that you can do this with complex songs where multiple instruments are played at once, specially when the same instrument playing multiple notes at the same time.)

Here's a good explanation why is it impossible what you ask (talks about MIDI, but the same stands for any tracker, TIC-80 patterns included).

Cheers,
bzt

[atesin]

If you're familiar with old-school trackers (FT2, OpenMPT, MilkyTracker etc.), then those work similarly like TIC-80

yes, but they do read and import .wav files as instruments, while tic80 dont

Another approach try using Audacity. Load a violin wave for example which is 1 second long, then on the bottom left enter "32 Hz" (this only works if the length of the sample is 1 sec). The resulting waveform should then look like exactly like the one you have to set in TIC-80 for a similar sound.

NICE HINT!

run a tool to read wav files, parses and put data in a .cart directly

Ah, about that, it won't work. You might be able to use wav files for the waveforms only, and not for the entire song as stored in the .cart.

hi.... i never talked about importing an entire .wav song inside a cart, that is unlikely... but about putting sound samples in sfx editor wave slots instead

i would like a tool to get some .wav SAMPLES of about 1 second, fixed frequency, etc.. and make it appear inside tic80 sfx editor wave slots (obviously with lower quality)..... that way we can get some samples from outside and make nice songs or sfx's in tic80

[bztsrc]

about putting sound samples in sfx editor wave slots instead

TBH I got so frustrated not getting answers to my TIC-80 audio questions for months that I wrote my own fantasy console in the meantime... It can import / export Amiga MOD (.mod) tracker files and General MIDI (.mid) music sheets too (for the latter there are some built-in wave samples or you can import .wav files directly). To see it in action click here (just drag'n'drop the "demo floppy" to the floppy drive, and the music will start).

i would like a tool to get some .wav SAMPLES of about 1 second, fixed frequency, etc..

The soundbank for the built-in waves for my fantasy console is here. It is in Amiga MOD format, but all you need is the samples in that file. Has 2 instruments per General MIDI family, each approx. 1 sec long, max 16k samples, all tuned to C-4 (except the violin, that's tuned to G-4). It is exactly what you asked for, but you'll need some work to scale these down to TIC-80 waves.

Oh, one more thing, these samples are licensed under GPL.

Hope this helps,
bzt

[joshgoebel]

I'm fairly certain that it does, at offset 0xFFE4 (see wiki).

Yes, if we're being quite technical... :-) but that (at a very slow cadence) is only 1/60 a second of wave data... there is no where clear to just import a 20 second (or even 1 second) wav data - which is what this user is requesting when they say "wav data"... they mean a few seconds (at least), not a 1/60s waveform.

i would like a tool to get some .wav SAMPLES of about 1 second

The TIC-80 "samples" (waveforms) are 1/60 a second. So a full second would require 60 samples, but we have only 16. So you're back to trying to copy things in and out of the audio registers if you want like 1 second samples.

You might first try downsampling your content to 2.1khz and see if you're even happy with the result... it sounds pretty awful. (I tried a whole song)

[atesin]

@joshgoebel : see the 16 waveform slots at sfx editor?

well... i wish some tool to import external waveforms to THAT slots, whatever the duration or quality they have ..... i never had talked about a 20secs song or whatever... you are making too many wrong asumptions about i said

if waveforms at sfx editor stores 1/60 second, then the .wav samples to import should have 1/60 second.... no problem

---

i hope i had talked clearly enough... i am thinking in something like, for example:

• process a dog barking sound in audacity or other software... with fixed frequency, duration, etc
• save as wave, doesnt matter if too short or low quality
• execute some tic80 command or external tool like ...

> import sfx woof.wav id=11

• then... it appears in bank 11 waveform, avaliable for our game or app like some 'dog game' or song

• additionally, you could later import a midi pattern to make a song using this new sfx... but that is outside the scope of this request

[joshgoebel]

I think you're not imagining how short 1/60 of a second is... 🙀 you couldn't put "woof" in there, but perhaps just part of a very rough w or o sound...

And I think my suggestion of sox is about as close as you're going to get...

• sox gets you the raw data in 8-bit binary form (ie wav or mp3 => raw 8-bit)
• read new file one byte at a time, downsample to 4-bit (shift right)
• output ASCII ready to cut and paste into cartridge external editor

ChatGPT could probably even help you with those last two. :)