16 users online:  bebn legg, Children's Digest 1950-2009, Daffster,  GlitchCat7, Hayashi Neru, margot, mathie, NewPointless,  NopeContest, nyankek, OEO6, Phyll, RZRider, solgaleo35, spooonsss, Toshi - Guests: 60 - Bots: 834 Users: 55,024 (2,219 active)
Latest user: biblicaldemoman
Not logged in.
Posts by nyanpasu64
nyanpasu64's Profile - Posts by nyanpasu64
Pages: « 1 2 3 4 5 »
Originally posted by MaxodeX
They represent note length in ticks. c=24 c8

WAIT that's a thing?!!!!!?!?!?!?! I'm totally using that in the future!
I added a "custom title" through CSS, is that not allowed?
Uh, your download link still links to the original AMK zip with addmusick.exe from 2017-04-21 7:39PM PST.

Files modified: /asm/SPC/include/S8A_SkipKeyOff.asm and /asm/SPC/include/S8_RemoteCommands.asm. Nothing else changed.

EDIT: What happened to 6646's AMK mod?
https://www.smwcentral.net/?p=viewthread&t=61465&page=43&pid=1398121#p1398121 https://github.com/boldowa/AddmusicK/tree/amki_dev
(on my port) I could've added echo... if I had freed up a few kilobytes worth of samples... But I realized some parts were missing and fixed them, less than 30min before the deadline...

My bass is missing quite a few pitch bends/vibrato. Oh well :(
I get a feeling that I can't submit a song that only inserts without global songs or SFX into the music section?
@jonkaruzu omg, I've spent a long time looking for SF2comp, sf2 to wav/text extractor and recompiler, after having forgotten the name. I'm bookmarking it.
probably won't participate, but is spc+vgmtrans=midi+???=txt allowed?
designing a midi->mml converter which outputs AMK-compatible instruments (with a mapping table), and (to some extent) volumes and pans is probably doable. would that reduce porting effort *too* much? (I guess not if we have to cut down channel count)
I'm trying to use the `addmusick -m -norom song.txt` command. This works on the April 2017 AMK 1.1.0 beta, but not the May 2017 beta where I get:
Error: File "music\" not found.

Regardless of what I put in song.txt.

Removing -m prevents this bug from occurring (but bloats the SPC file size).

I'd prefer not to put my song names in Trackmusic_list.txt, since that file lists *all* songs to be compiled at once, stored with the AddmusicK program (instead of storing the .txt name with my build scripts). If I want to switch songs, I have to change a text file located in the AMK program directory, instead of just running a different build script.
Since everyone length is at least 4, that's literally the same thing as setting EVOL[LR] = MVOL[LR] * feedback/0x80 (or 0x7f?). It's still just as much echo, just less controllable. Everything gets sent through fir at least once, and if feedback is zero, no extra echo is audible.
6.89/10, I'm not sure I like the character design. pixel art seems fine. Are the angular (directional) edges intentional?

"""spcplay is most accurate""" That is incorrect.

- foo_gep is 100% accurate, since it's based off Higan's slow but hardware-accurate SPC emulator core (cycle-accurate and sample-accurate down to the last bit).

- SPCPlay is based off a faster codebase with more non-hardware twiddle-knobs and a less-accurate track record.
- A few years ago, I actually reported a pitch modulation bug to the author, who *increased* the strength to match foo_gep and SNES hardware. (Kefka's laugh wasn't rough enough in spcplay).

Recommending people use spcplay as a reference point for how things *should* sound is dangerous misinformation.


foo_gep has a few twiddle-knobs, which should be set this way for hardware-accuracy:

- Remove surround effect (disable)
- Interpolation: (Gaussian)

It hides several extra twiddle-knobs under Advanced/Playback/Game Emu Player/SPC:
- Enable Blargg's analog filter simulation (enable)
- Enable 1.4x floating point gain scale (don't know, disable)
- Enable optional oversampling by setting a count above 1 (1)
- Enable SPC echo emulation. (enable)


Download at: amktools.7z

wav2brr is a tool to automatically batch-encode many .wav files to .brr files, loop at given loop points, resample by ratios, and calculate tuning for the final sample.


Just create a folder named “wav” (or anything else), and place each .wav (along with Audacity projects, etc) in its own folder (name doesn’t matter).

For example, in the folder “wav/Organ” with file “organ.wav”, create file “organ.cfg” containing


Decrease ratio to decrease the size of the resulting .brr file (tuning will still be correct).

loop and truncate describe the begin and end of the looped region. Your sample will be automatically resampled to a multiple of 16 samples, which is very convenient.

The generated tuning will be correct if your sample is tuned to an exact note, and at contains the correct MIDI pitch.

wav2brr also supports extracting looping and tuning information from .sf2 files. Pass --sf2 <sf2path> to wav2brr.exe, and remove loop, truncate, and at.


Assume your samples are located under wav, AddMusicK is located at amk_path, and you want your samples to be written to samples/smp.

Create file convert_brr.cmd containing:

path\to\amktools\wav2brr.exe wav amk_path smp

See https://github.com/jimbo1qaz/amktools/blob/master/docs/wav2brr.md for full documentation.

Importing tuning data

wav2brr prints tuning data to console. To automatically insert tuning data into .txt, see my tutorial at docs/mmkparser.md .


  • mmkparser will interpret all lowercase v and y as volume and pan commands (which can be rescaled). If you map "octave=o3" and use octave, mmkparser will report an invalid v command.
    • Potential solution: I could turn off v and p processing entirely.
  • In #instruments, if you include %tune followed by ;comments, the tuning bytes will be written after the comments. This is a bug.
  • In #instruments, if you use %tune, you cannot include more than one “sample.brr” per line.


wav2brr is based on brr_encoder (https://github.com/Optiroc/BRRtools).

I'm a bit concerned that you're mixing web-dev concepts from decades apart.

<center> was deprecated in HTML4 (1997).

CSS animation was first used in 2009, commonly supported since 2011, CSS3 standardization was in... uh dunno, the CSS spec has been divided into modules which standardize independently.
Astolfo: https://steamuserimages-a.akamaihd.net/ugc/838083785051395085/81FC2EBDDCCFFE0EA13C9CAB4B68FA2F33F6A946/
If I find a way to transform a sample into a series of single-cycle loops, then change the sample's loop point at 30fps (to compress voice), would the resulting waveform be legal? It is effectively a slightly shuffled series of single cycles with identical timbre as the original sample, but using around 1/10th the space (allowing for better treble reproduction than most snes vocals).
>30Hz is way too little to make timbre changes sound natural
i honestly believe I can produce good (not perfect, but good) results. My technique is inspired by the Famicom's N163 wavetable chip, which I have an earlier demo here: https://www.youtube.com/watch?v=nxAo2jA6GVs

The vowels are decent and have good treble response (although the transitions will not be completely smooth), the sibilants will be less-than-great, there's no reverb (but I can save ARAM and dedicate it to the echo buffer).
(ultima your avatars are qt... at least most of them)

Seems nobody else is interested in pushing technology to its limits... Nobody even commented about the quality of my Famicom vocal video :(

I could release my SPC code for other competitors in Idol9...
- WAV/BRR wavetable extraction is done.
- I haven't implemented syntax to play wavetables from an AMK .mmk/.txt file yet. (probably not until a few days later)
- I can get 5 seconds of vocals in 5-10KB depending on sample rate and framerate. The limiting factor is 255 wavetables+instruments per .spc file.

Once I finish my code, should someone/I host a SPCocaloid competition?
I think SPC interpolation is honestly not as bad as people make it out. If you use brr_encoder -g, it almost but not completely cancels the filtering (maybe even probably better than linear interpolation). It only seriously muffles your samples if you use bad tools.
Are we allowed to edit the loop points of the samples (only use part of it, like a single period of the waveform)? Begin *or* end?
Pages: « 1 2 3 4 5 »
nyanpasu64's Profile - Posts by nyanpasu64