Release

Midi Tools 1.15.5

SP4CEBAR 2021-05-02 20:12 (Edited)

click here to read what's new in version 1.15.5

This program can read, play, and convert MIDI files.

check out the API

You can choose to use the internal or an external disk, or both, but that's not supported yet. You can also choose what ROM entry to use. In this version, you can load midi files of 32KB and below, in a future version I want to increase this to 64KB (with two disks).

Loading A MIDI File

To load in a midi file, you can open a .mid file in a hex editor, if you're on IOS you could use an online hex editor*. In the hex editor, you can copy the hex data and paste it in this program, or the disk.nx program, or another program which can use this program as a tool. I've loaded a MIDI file for the Nyan Cat song in local ROM entry 5.

HEX Editor

* This online hex editor is one of the few that allows you to copy the hex data to your clipboard
Make sure to uncheck the "Use 0x and comma as separator".

Converter

When converting a midi file to an NX music file you can set the precision of the quantization: a value of 1 means that it'll preserve the music as best as it can, but it'll probably not fit in the 64 tracks of NX music, and the playback speed could be very slow.

Loading 64 KB

Although the program doesn't support it yet, you can theoretically load 64KB using both the internal and an external disk. To do this you'll need to split the file into two files of 32KB or less. I think the best way to do this is to use a text editor with numbered lines like this compiler website. Then paste the 64KB of midi data into the text editor, go to the end of line 2048, and hit enter. Copy the first part to the disk.nx program or another program that can use this program as a tool, make sure it doesn't have any other ROM file in it. Finally, copy the second part to the ROM of this program, and remove/replace the midi data stored in ROM 5.

Loading From Strings

Since version 1.15.1 you can load MIDI files from strings of hexadecimal characters (without spaces or newlines). These strings have to be written inside the G_FILE_STRING1$ variable inside the GET_FILE_STR subprogram (which is the last subprogram). This method allows you to load infinitely big files. Unfortunately, you wouldn't be able to post any program larger than 128KB, luckily you can use the API to scrape this data and store it as a new file.

Versions:

1.15.5 - QUICK PATCH
1.15.4 - DEBUGGING
1.15.3 - EFFICIENCY
1.15.2 - DEBUGGING TOOLS
1.15.1 - STRINGS ARE SUPPORTED
1.15.0 - SCRAPE HUGE FILES
1.14.0 - CLEANUP CONCLUDED
1.13.8 - "I" INFESTATION CLEARED
1.13.7 - NO MORE J
1.13.6 - ROGUE PRINTS OBLITERATED
1.13.5 - MIDI HEADER REVISIT
1.13.4 - CLEAN UP
1.13.3 - OLD MODES MERGED
1.13.2 - FAST TRANSCODING
1.13.1 - SMOOTH ANIMATIONS
1.13 - THE API IS REAL!
1.12.6 - IT'S TIME TO MERGE
1.12.5 - CLEAN API
1.12.4 - CLEAR GLOBALS
1.12.3 - CORE SEPARATED
1.12.2 - LOADER SEPARATED
1.12.1 - CORE REFACTORED
1.12 - LOADER REFACTORED
1.11 - NOTE RANGE BOOST
1.10 - NO MORE HANGING NOTES
1.9.1 - EMBEDDED DATA
1.9 - DISPLAYS AND AUTOMATION
1.8 - NO MORE FLICKERING
1.7 - CONVERTOR IS BACK!
1.6 - ACCURACY!
1.5 - RELIABILITY!
1.4 - MULTITRACKS!
1.3.1 - EFFICIENT VOICES
1.3 - BIG FILES
1.2 - RUNNING STATUS
1.1 - EXISTENCE

Known Bugs And Errors:

• Some MIDI files spam a bunch of controller change commands in between the notes, which in the current version causes the convertor to not always write a playing note down.
• Aftertouch isn't supported, and it might cause some high-pitched beeps, or the beeps may be unrelated to aftertouch (or it ma be fixed by now)
• Hanging notes... they happen and I have no clue why
• Tacks aren't played all at once when they should, I should look into this more (no clue what this means, I probably fixed it by now)
• Many songs are played too fast because the midi timing definitions are complicated and I didn't properly implement them.

With each update, the program changes quite a bit. So If something doesn't work, just try a previous version.

Big thanks to Timo for helping me with the midi file system

CubicleHead 2021-05-03 11:30

So awesome! This will be very useful!

SP4CEBAR 2021-05-03 18:10

Thanks

G-9 2021-05-04 12:07 (Edited)

Still not working :(
EDIT : I tried ur file to hex converter and it worked :3

Disk.nx | Open in app

2021-05-04 12:31

SP4CEBAR 2021-05-04 14:17

Great

G-9 2021-05-04 18:54

Nice markdown :3

SP4CEBAR 2021-05-04 21:26 (Edited)

Thanks
When I realized that markdown is the same markup language discord uses, I could use it straightaway

SP4CEBAR 2022-01-21 16:24

Read, play, and convert used to share the same loop, Man was that annoying to work with
It was one enormous block of code

I've just broken it up into subprograms and separate loops for each operation, it looks so much cleaner now

SP4CEBAR 2022-01-21 17:32 (Edited)

Also I used the variable "I" to hold the memory address, and whats worse is that I made it a global variable, why did I think that was a good idea

SP4CEBAR 2022-01-21 17:43

I managed to change the order of the tasks
It now reads the next delta T after it has executed a midi command (instead of before)
With this I know the time to wait in advance, which is essential if I want to make a scheduler (multitask)

G-9 2022-01-22 14:41

Amazing you did a real great job
congrats

SP4CEBAR 2022-01-22 16:54

Thank you!

SP4CEBAR 2022-01-22 21:22 (Edited)

I think I figured out why some tracks won't play correctly
I think that storing the address and the timecodes for each track in the scheduler isn't enough, I think my decoder needs more variables (it's been a while since I last worked on it)

SP4CEBAR 2022-01-22 21:28 (Edited)

I caught my midi decoder jumping $800 addresses ahead, so clearly it's missing something...

SP4CEBAR 2022-01-22 22:50 (Edited)

The jumping is caused by it reading a metadata command with a really long length, which is wrong
So it must've gone off-track earlier...

G-9 2022-01-23 14:36

I'm so hyped for both bugfix and multitrack convert :D
The song output would be amazing 🤩
Also to you have a trick about how to guess precision number ? My #15 file is filled with A-0 😂

SP4CEBAR 2022-01-23 17:51

I'm trying to make it play multiple tracks at once, this requires a scheduler and things get more complicated, I'm currently filling my program with tracers to compare the multitrack player with the single track player
The converter doesn't need a scheduler, it converts the tracks one by one (NX tracks are monophonic)
The precision is used for time quantization (aligning notes to a grid), when the NX file size is too big, you may want it to be less precise, and if it doesn't sound as good, you may want more precision (you may not be able to fit the whole song inside the 64 NX tracks)

(unstable wifi connection made me have to rewrite this)

SP4CEBAR 2022-01-23 19:33

I just got an idea: I could make a new file format called "NX Tracks" which contains just the tracks part of the NX music file format, but I can store as much NX tracks as needed
I could make my music editor compatible with that format as well
It would allow the converter of this midi tool to make music files that are much bigger

SP4CEBAR 2022-01-23 21:02

I have this bug memorial in my code:
'BUG MEMORIAL: FORGOT TO OFFSET THE INDEX VALUES
I think it just happened again...

SP4CEBAR 2022-01-23 21:04

The first 8 bytes of each midi track contains the size of the track, which shouldn't be read as midi data, because it's not
This is what a good offset looks like: FOR I=INDEX(J)+8 TO INDEX(J+1)-1

SP4CEBAR 2022-01-23 21:06

YAAAS, the multitrack reader can now read the nyan cat midi file (which it couldn't before)

SP4CEBAR 2022-01-23 21:09

AND THE GLITCHY NOISES IN THE ZELDA THEME ARE GONE!!!

SP4CEBAR 2022-01-23 21:28

And the track playback is in sync!

SP4CEBAR 2022-01-23 21:29

It's done!!
*Well at least the scheduler-based midi player*

SP4CEBAR 2022-01-23 21:30

Next I'll try to find a way to optimize the usage of the four voices

SP4CEBAR 2022-01-26 21:18 (Edited)

The tracks slowly drift out of sync
This is probably caused by lag (the program can run behind on schedule), and the rounding of delta times (I have to convert delta times to NX-frames)
That's the issue with delta times, they are relative, it can't know whether or not it's synced with the other tracks unless it has an absolute time count to compare it to.
So I guess I've got to sum the delta times together for each track so that I can convert that to NX-frames and subtract TIMER from it to generate synced delta time values.

SP4CEBAR 2022-01-26 21:26 (Edited)

Also I think the converter needs more options
Currently you can only convert the tracks from the start until the memory (NX tracks) runs out
I want to add an option to select a single track to convert
Also, I want to add an option to export quantized notes as a file
And, I want to add an option to export unquantized notes as a file
I want to make my piano roll music studio compatible with those files so that you can do more post processing to improve the quantization quality
But it's a ton of work I'll probably not do any of this soon

nathanielbabiak 2022-06-05 03:52 (Edited)

Hey, as you were developing this, did you learn anything about the audio registers? I posted a question here... Any chance you could take a look?

SP4CEBAR 2022-06-05 20:15 (Edited)

@nathanielbabiak yes, I've done some tests myself on some of the audio registers, I'll try to answer your questions
I remember that I once understood how sounds are started and stopped via the registers, but I may have forgotten it now, I'll see if I can find some code that starts or stops sounds via registers
Either way, this program displays the state of those registers, and that was quite helpful for my research

SP4CEBAR 2022-06-05 20:20

I found a program that writes registers to start playing an arpeggio when you touch the screen and the sounds stop once you stop touching the screen

Voice Simulator.nx | Open in app

2022-06-05 20:20

SP4CEBAR 2022-06-05 20:29

After analyzing my code, I think this starts a sound
POKE $FF42,PEEK($FF42)MOD 64+64+128
And this stops a sound
POKE $FF42,PEEK($FF42)MOD 64+64

And there's also this, with the comment "INTITALIZE VOICES" above it
'POKE $FF42,PEEK($FF42)MOD 64+128

SP4CEBAR 2022-06-05 20:36

These lines of code read the status byte of voice 1 (peek($FF42)), set the most significant two bits to zero (mod 64), and then add their own most significant bit 2-bit codes (+64) and or (+128)

SP4CEBAR 2022-06-05 21:16

I just posted an answer to your question
I guess I wrote you a textbook

nathanielbabiak 2022-06-06 03:38

Thanks - this is excellent stuff!

SP4CEBAR 2022-07-31 10:39 (Edited)

Version history (edit)

1.2 - important fix

• Running status is now supported! Most MIDI files should work now!
• Some midi files don't use key-up commands, instead, they use key-down commands with a velocity of 0, which caused some problems, luckily That's fixed now.

1.3 - Never gonna give you up

• You can now load files up to 32Kb!
• Chords are now supported in the midi player. (Limited to four voices)
  1.3.1: I made the tones rotate over the four voices, to use them as much as possible

1.4 - MULTITRACKS!

• I added a scheduler so that it can play all tracks at once, you can access it with mode "M". It's still an alpha version, so the sound is a bit messy and not all songs will work. You can load Local 4 in this demo
• I cleaned up the code, now it's actually doable to work with it

1.5 - RELIABILITY!

• My new schedule-based MIDI player works reliably now! For songs that have their instruments divided up into multiple tracks, this means that you can now play all tracks at once, and you can finally get the full experience of the music. I'm really happy now! :D

V1.6 - Accurate Playback

When playing a MIDI file, tracks don't get out of sync anymore!
Even single-track playback has improved, it's much more accurate now!

I used to read each delta time value, round it to NX frames, and store it, this results in some tracks being rounded upwards more and other tracks being rounded downwards more, and thus the tracks get out of sync

The solution was really simple, I didn't need to make an absolute schedule, I just needed to store and sum the delta time values, and round them to NX frames at the very last moment, I also needed to store a TIMER_ZERO value so that I could use TIMER for this.
this is a self-correcting system, the sum of delta time values is the exact right moment, no matter what time the program takes to run it
return to top

SP4CEBAR 2022-09-21 20:12 (Edited)

V1.7 - Conversion Is back

I've been so focused on multi-track playback that I didn't realize that conversion was broken, and as it turns out it was the lack of testing that broke it
I found this:
IF NOT(STATUS>$7F AND STATUS<$A0) THEN
Which should've been:
IF STATUS>$7F AND STATUS<$A0 THEN
I must've not tested it, otherwise I would've noticed

However, it seems like for some files you need to enter a really high precision value (a high precision value means low precision) like 32 or 64
return to top

McPepic 2022-09-22 00:29

@SP4CEBAR
I was messing around with different files and I found one that temporarily breaks the LowRes display when played using the tool. There's also some weird stuff going on with the audio. Would you like to have a look?

SP4CEBAR 2022-09-22 08:26 (Edited)

Yeah, sure
does the display flicker? If so then that song probably has a ton of tracks that are played so fast that VBL has to interrupt the graphics, and the four voices are having a hard time...
Ohh, I finally know a solution to the flickering: I shouldn't use CLW
instead, it should just relocate the cursor, and print blank lines when needed

SP4CEBAR 2022-09-22 08:49 (Edited)

V1.8 - No More Flickering on busy MIDI files

• CLW has been replaced with LOCATE 0,0 and a BG FILL for the lower half of the screen, and some extra spaces have been added to "PRINT"s to overwrite all of the underlying text
  return to top

SP4CEBAR 2022-09-22 10:05 (Edited)

V1.9 - Displays And Automation

• MIDI track display on top
• NX voice display below
• instant playback mode, to activate it:
  • open the source code in a code editor, look at the first lines
  • set the AUTOMATIC_PLAYBACK variable to 1

Also... the monkey island midi track, which played OK before, now sounds broken...
return to top

SP4CEBAR 2022-09-22 10:15 (Edited)

V1.9.1 - Embedded Data

• when big local files are loaded, NX can't load it's characters anymore, so these are now moved into a "DATA" block in the code
  return to top

McPepic 2022-09-22 15:17

Wow, you’ve been really busy. Super cool program btw! :)

SP4CEBAR 2022-09-22 15:39 (Edited)

Thanks! the last two updates were basically a merge with the instant midi players like the monkey island midi player, this was needed because it's annoying to have two versions that both need updating

Also, Did the converter work this time?
Does the file that looked like it had graphical glitches look better now?

McPepic 2022-09-22 17:20

So, I was messing around with the program some more. After the file plays for a while, it gets stuck on a really high pitched beep on voice 4. I don't know if it ever stops this.

If I try to go to the graphics / sound editor for LowRes, it gives me a black screen. If I try to run the program again from the menu, it gives me the default blue screen of a blank program. After a few seconds, the program finally loads, though.

The converter plays back the result (I find this program works best with a precision value of 64). When I open the music editor, the patterns are all set to track 0 on every voice and the sounds are all set to 0s on every option. The tracks do seem to be filled in properly by the converter, though. Also, the music is always saved to file 15 of the ROM regardless of what you entered in for the conversion. This seems to be hard coded.

Midi Tools Organized (2).nx | Open in app

2022-09-22 17:20

SP4CEBAR 2022-09-22 22:54 (Edited)

The stuck beep is probably the same problem that the Monkey Island theme had: for some reason, the program sometimes won't stop a note in running-status files

It's not the program, it's the file that determines what precision value is needed, for example, the midi file at local ROM(5) works with a precision of 2

Yeah, the patterns aren't given any data as the convertor in its current state is pretty limited, it just extracts pitch values and encodes it into tracks

Good that you found that ROM 15 can't be changed, I never realized it, so I must've completely forgotten it

SP4CEBAR 2022-09-23 08:46 (Edited)

It looks like the stuck beep, got missed halfway through the song, and the song happens to not overwrite it (it uses one voice)
The beep has an NX pitch of 117, an NX velocity of 1, and a track of (7 and 16) (midi has tracks (tags) within tracks (data blocks))

SP4CEBAR 2022-10-29 13:00

I caught the hanging note bug!

SP4CEBAR 2022-10-29 13:02

And I found some suspicious code:

  FOR X=0 TO 3
    IF PLAYING(0,X)=PITCH THEN 
      VOI=X
      EXIT
    ELSE IF PLAYING(0,X)=0 THEN 
      VOI=X
      EXIT
    END IF
  NEXT X

SP4CEBAR 2022-10-29 13:03

When PLAYING(0,0)=0
it'll never be able to select PLAYING(0,1)=PITCH to be able to overwrite it

SP4CEBAR 2022-10-29 13:07

It's fixed, it was so simple

SP4CEBAR 2022-10-29 13:15 (Edited)

version 1.10

• No more hanging notes on running status files that use key-up with velocity=0 as key-down!

bugs remaining:

• controller change can freeze the scheduler (music freezes as long as it's active), like when your computer (windows scheduler) freezes because of one program, this should be an obvious fix
  return to top

SP4CEBAR 2022-10-29 13:53 (Edited)

monkey island still has hanging notes, I think the issue there is that it doesn't support enough midi commands

SP4CEBAR 2022-10-29 14:52 (Edited)

Next I'll replace PLAY V,N with this function, to get more range:

SUB PLAY_(V,N,O)
  A=$FF40+12*V
  POKEW A,O+16*16.35*2^((N-1)/12)
  POKE A+2,PEEK(A+2) OR %11000000
END SUB

SP4CEBAR 2022-10-29 15:17 (Edited)

Version 1.11

• more range of notes, really low notes and really high notes can now be played
  return to top

SP4CEBAR 2023-09-09 14:47 (Edited)

I'm in the process of refactoring the code so that I can make an API (which I will probably post as a new program). Anyway, when I made this MIDI program I would copy and paste code rather than make functions for it. I just found an inconsistency between these lines of the note scheduler:

• TRACKD(J,0)=TIMER-(DT3>0)*DT3
• TRACKD(J,0)=TIMER-(DT3>=0)*DT3

Addressing this inconsistency in the code may fix some bugs, like the hanging notes on more complicated midi files

SP4CEBAR (formatted to look like a new post)

EDIT: It didn't fix anything unfortunately

SP4CEBAR 2023-09-09 20:07 (Edited)

Version 1.12 - LOADER REFACTORED

• Most of the code has been refactored. To turn this into an API, I'd have to separate the visualizers, the user input, and the midi processing. I've tested the midi reading and playing modes, but I haven't tested if the converter still works.
  return to top

SP4CEBAR 2023-09-11 08:55 (Edited)

Version 1.12.1 - CORE REFACTORED

• the core interpreter has been refactored
  return to top

SP4CEBAR 2023-09-11 16:19 (Edited)

Version 1.12.2 - LOADER SEPARATED

• The file preparation has been separated into midi interpretation, printing, and user input
  return to top

SP4CEBAR 2023-09-13 17:32 (Edited)

Version 1.12.3 - CORE SEPARATED

• refactored the code even further, especially the code of the midi code interpreter
• separated out all the prints in the midi multitrack loop
• I still haven't tested the file converting, so there's a decent chance that it's broken by now
  return to top

SP4CEBAR 2023-09-16 07:55 (Edited)

Version 1.12.4 - CLEAR GLOBALS

• prefixed all global variables but the fundamental, short, and really hard to replace, S,I,V,V0,V1,V2, and V3 with "G_" to make it easier to spot global variables in the code
  return to top

SP4CEBAR 2023-09-29 13:40 (Edited)

Version 1.12.5 - CLEAN API

(the NX file is accidentally named "1.12.3")

• renamed more 1-letter variables: (V and S)
• made unnecessary API function parameters global
• roughly refactored the conversion function (it works, but the generated file sounds a bit off)
• moved the "end of file" indicator inside the printing function so that it can be disabled
  return to top

SP4CEBAR 2023-10-04 09:26 (Edited)

Version 1.12.6 - IT'S TIME TO MERGE

• the midi interpreting and note scheduling are now separated from the actions: like waiting for the next note, playing the note, and printing information on the screen.
  This is really useful as it allows many functions to be merged: the MIDI converters don't need to wait for notes and play and print them, so the converter can just call the aforementioned MIDI interpreting and note scheduling function. This removes duplicate code and makes this program easier for me to develop.
  return to top

SP4CEBAR 2023-10-04 15:05 (Edited)

Version 1.13 - THE API IS REAL!

• new peak meter display
• the MIDI interpreter has been cleaned up further:
  • the function that updates the scheduler with newly read delta times is now called at the beginning of the MIDI playback loop, this allowed a function to be removed from the playback initialization.
  • more variables are made global so that they don't have to be passed around and chained from function to function
• MIDI_SCRAPE function that loads MIDI data in time-chronological order into three arrays: TIMES, TONES, AND TRACKS. Currently, the scraping process takes as long as the playback takes because the MIDI interpreter is still dependent on NX's TIMER variable, this will change in a future version.
  return to top

nathanielbabiak 2023-10-04 21:30

These latest improvements make the code really clean!

SP4CEBAR 2023-10-05 07:09

thank you!
I found that organizing code and merging subprograms is really satisfying to do

SP4CEBAR 2023-10-07 08:54 (Edited)

Version 1.13.1 - SMOOTH ANIMATIONS

• animations (like the peak display demo) are now called every frame, making them look way smoother.
  return to top

SP4CEBAR 2023-10-07 11:37 (Edited)

Version 1.13.2 - FAST TRANSCODING

• The MIDI algorithm is now independent of NX's TIMER variable. Now it can be sped up and used for transcoding. The MIDI scraper takes only a few seconds now to scrape the notes from a MIDI file.
  return to top

McPepic 2023-10-07 13:52

What does it mean if I get:
Variable not initialized
V0=V0+256*V
?

SP4CEBAR 2023-10-07 14:34 (Edited)

It means that your file has a MIDI "E" command and that I missed a V when replacing all occurrences of "V" with G_BYTE (where "G_" means that it's global). The line should've been:
V0=V0+256*G_BYTE
Thanks for reporting this by the way!

SP4CEBAR 2023-10-07 15:17 (Edited)

Version 1.13.3 - OLD MODES MERGED

• crash in midi E commands fixed.
• playing single tracks and reading midi are now better refactored, and separated by function and the two are merged.
  return to top

SP4CEBAR 2023-10-07 22:24 (Edited)

Version 1.13.4 - CLEAN UP

• removed the remaining duplicate midi reading subprograms after the merge with MIDI_PLAY_SINGLE.
• organized all subprograms by labeled categories.
• organized some code like the refactoring of the functions that define the data for the graphics registers (without using ROM).
• fixed the delta time for the original display display of the multitrack player.
• the single track player now uses the same printing function as the multitrack player.
  return to top

SP4CEBAR 2023-10-08 12:47 (Edited)

Version 1.13.5 - MIDI HEADER REVISIT

• the API commands at the top have been wrapped inside a function to make it clearer and reduce the number of comments.
• the MIDI file header specifies three types of files: 0: one track, 1: simultaneous tracks of which the first contains some special timing info, and 2: consecutive tracks. Of these modes, only mode 1 needs the tracks to be played all at once, so I redirected the other ones to the linear player.
• MIDI files are usually played too fast by this interpreter, so I've looked some more into MIDI headers, I've made a bit of progress, but it is quite complicated, currently I've just done a bit of the time definitions in the header:
  • the MIDI header has one of two definitions for the delta times:
    • "ticks per quarter note", I'm not sure how long a "quarter note" is meant to last.
    • "ticks per frame" and "frames per second", which, when multiplied together would give you "ticks per second".
  • the MIDI file could contain meta events that could be used for some additional special timing information
    return to top

SP4CEBAR 2023-10-09 11:00 (Edited)

I think I could catch the hardest bugs by exporting the arrays of the MIDI SCRAPER to the disk and then comparing it to the output of a MIDI API in a programming language like Python

Funnily enough, I wanted to learn Python before I started programming (in LowRes Coder), but now I've basically learned and used every programming language but Python.

SP4CEBAR 2023-10-09 13:16 (Edited)

Version 1.13.6 - ROGUE PRINTS OBLITERATED

• "no file" detection so that it doesn't throw an error if there's no MIDI file at the specified location.
• removal of rogue PRINT statements from specific MIDI commands, because all prints should go through one of the MIDI visualizers.
• more variables have been made global.
• global variables have been removed from the MIDI VISUALIZER subprogram's IO
• more variables have been renamed to remove abbreviations and make them clearer.
• the API and mode control section at the top of the program has been made cleaner by the reduction of comments, and the renaming of functions
  • "GRAPHICS" has been renamed to "VBL_GRAPHICS"
    return to top

SP4CEBAR 2023-10-09 15:37 (Edited)

Version 1.13.7 - NO MORE J

• fixed a typo: "SYNC" instead of "SYNQ"
• renamed all occurrences of "J" to "G_CURRENT_TRACK", except the occurrences where J doesn't mean CURRENT_TRACK, also G_CURRENT_TRACK has been made global.
• the SUB MIDI_VISUALIZERS with its CALL PRINT_MIDI_TOOLS, now have no more parameters: all of them have been clarified, prefixed with "G_", and made global
  return to top

SP4CEBAR 2023-10-09 16:27 (Edited)

Version 1.13.8 - "I" INFESTATION CLEARED

• replaced (probably) all occurrences of "I" with G_ADDRESS, now "I" is no longer a global variable, and new FOR loops (with I) won't cause bugs anymore.
  return to top

SP4CEBAR 2023-10-09 19:15 (Edited)

Version 1.14.0 - CLEANUP CONCLUDED

• Unnecessary parameters have been removed from functions, this is possible because most common variables within this program are global.
• Global variable resetting has been extracted to separate functions.
• This should conclude the clean-up of this program
  return to top

SP4CEBAR 2024-01-07 00:18 (Edited)

Version 1.15.0 - SCRAPE HUGE FILES

• Location parameter "S" will load the string provided in the GET_FILE_STR subprogram as a MIDI file (currently only accessible with API functions), you can do anything with these, including MIDI scraping
• Instead of playing scraped MIDI files, you can now export them as a file of interlaced arrays:
  • Every four bytes are: times array (1 byte), tones array (2 byte (POKEW / PEEKW)), tracks array (1 byte)
• You can use the ARRAY_PREPARE_FILE subprogram, which works like the MIDI_PREPARE_FILE subprogram, but loads interlaced array files that contain scraped MIDI, and only supports "L" (local) and "E" (external).
• In the ARRAY_VISUALIZERS subprogram you can turn off (comment-out) the visualizer of scraped MIDI playbacks.
  return to top

SP4CEBAR 2024-01-12 15:47 (Edited)

Version 1.15.1 - STRINGS ARE SUPPORTED

• Cleaned up some of the mess around the scraped MIDI subprograms in the API
• loading from strings is now supported in the MIDI tools CLI
• the locations $A000 and $4000 in the code have been replaced with global variables (with the same values of course)
  return to top

SP4CEBAR 2024-01-12 19:36 (Edited)

Version 1.15.2 - DEBUGGING TOOLS

• G_CACHE_SIZE has been reduced from $4000 to $1, this should enhance the performance as most files will have multiple tracks playing at the same time that lie in different cache sections.
• the MIDI_SCHEDULER_INFO subprogram has been improved, it now has addresses that start at zero and a display of how much time is left until the next command of each track
• I've removed some abbreviations within subprograms and I've removed some variables that were recently made global from certain function calls (functions don't need to pass along global variables)
  return to top

SP4CEBAR 2024-01-12 22:26

Version 1.15.3 - EFFICIENCY

• ROM files are read straight from ROM without any cache.
• it is more efficient to run the MIDI graphics on VBL than after each MIDI command so it now runs on VBL
• the MIDI scheduler info now displays the MIDI command type of each track as well
  return to top

McPepic 2024-01-12 22:39

Awesome! I just got it to play the Tetris music and it sounds great. Good work @SP4CEBAR!

SP4CEBAR 2024-01-13 08:30

Thank you!

SP4CEBAR 2024-01-24 16:15 (Edited)

Version 1.15.4 - DEBUGGING

• The sequencer display has better colors for the numbers now
  • red for a key-down
  • the MIDI code is gray when the track stops just like the address is
• the error that pops up when your file has more tracks than the arrays can handle is now caught and displayed on the screen
• if the cache loader ends up peeking outside of working RAM (for whatever reason), this error will be caught and displayed
• The cache loader was in a horrible condition, now it's in a patched up horrible condition, I fixed and changed:
  • when loading from the API the global variables G_OPERATION$ and G_FILE_IN wouldn't be set, now they would. (I wonder if there are more variables like that with the same issue)
  • even though I thought the cache refreshing function would have be able to refresh multiple cache segments, it wasn't working like that: it would keep on reading, even outside of the working RAM without refreshing the cache (and somehow it still worked without any problems). Now I fixed that and it now does refresh the cache. The problem is that big files have tracks in multiple cache segments. In order to play those tracks at the same time, the program has to refresh the cache many times. The problem with that is that the LOAD command is really slow, more information here
    return to top

McPepic 2024-01-24 16:40

Argument count mismatch
SUB MIDI_QUANTIZE( G_FILE_OUT )

SP4CEBAR 2024-01-24 18:22 (Edited)

that one must've happened during the clean-up of the code, it has likely been there for a long time (I haven't touched or tested the quantization in a long time)

SP4CEBAR 2024-01-24 18:37

Version 1.15.5 - QUICK PATCH

• the MIDI quantizer no longer gets an argument count mismatch when called from the MIDI tools GUI
  return to top