https://github.com/sevagh/drum_machine

create click tracks from harmonixset annotations on the fly with libmetro, libsoundio, and stk
https://github.com/sevagh/drum_machine

ast beats click-track lexer-parser metronome music-information-retrieval rhythm tokenizer

Last synced: 12 days ago
JSON representation

create click tracks from harmonixset annotations on the fly with libmetro, libsoundio, and stk

• Host: GitHub
• URL: https://github.com/sevagh/drum_machine
• Owner: sevagh
• License: mit
• Created: 2020-06-27T15:59:26.000Z (over 4 years ago)
• Default Branch: master
• Last Pushed: 2020-06-28T15:19:31.000Z (over 4 years ago)
• Last Synced: 2024-12-21T18:05:54.813Z (14 days ago)
• Topics: ast, beats, click-track, lexer-parser, metronome, music-information-retrieval, rhythm, tokenizer
• Language: Go
• Homepage:
• Size: 1.92 MB
• Stars: 1
• Watchers: 2
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        
Pipe [harmonixset](https://github.com/urinieto/harmonixset) beat txt files through drum_machine to listen to a clicktrack.

The file format drum_machine supports are the beat annotations here: [harmonixset/datasets/beats_and_downbeats](https://github.com/urinieto/harmonixset/tree/master/dataset/beats_and_downbeats)

The clicktrack is generated using [libmetro](https://github.com/sevagh/libmetro). The beat txt files are parsed using a handwritten tokenizer/lexer and parser, based on [this tutorial](https://blog.gopheracademy.com/advent-2014/parsers-lexers/). drum_machine is fuzz tested with the entire harmonixset dataset as a corpus, and fuzzing is done with [go-fuzz](https://github.com/dvyukov/go-fuzz) ([tutorial](https://dgraph.io/blog/post/continuous-fuzzing-with-go/)). Run fuzz test with `./fuzz_test.sh`.

### Mapping harmonix to libmetro

libmetro provides the following constructs:

* _beats_ are represented with _note_ objects, which are wrappers around a vector of floats generated from Stk. libmetro is more flexible but the C wrapper and use in drum_machine are limited to the MIDI Drum instrument, and 2 'timbres' of beats

* _bars_ are represented with _measure_ objects, which are collections of a sequence of _notes_ (can be padded with silence)

* the _tempo_ of the song is configured on the _metronome_ object, which can register a sequence of _measures_ to loop through

Although the result is not mind-blowingly exciting given the tempo and beat structure of pop songs is stable, drum_machine still a fun combination of several projects.

### Compile

Getting `drum_machine` built is not straightforward. You have to statically compile and install several C++ libraries from scratch. They must be compiled statically Go + cgo doesn't care about your configured dynamic library paths.

* https://github.com/andrewrk/libsoundio

After ensuring you have the correct dependencies (I've tested it only on Linux with ALSA, pulseaudio, and JACK libraries installed):

```

libsoundio $ mkdir -p build && cd build && cmake .. && make && sudo make install

```

* https://github.com/thestk/stk

I suggest you copy `rawwaves` to `~/rawwaves` and compile this way:

```

stk $ autoconf

stk $ ./configure

stk $ ./configure RAWWAVE_PATH='$(HOME)/rawwaves/' BUILD_STATIC='yes'

stk $ cd src && make && sudo make install

# manually install the .a file, it's not in the makefile i believe

stk $ sudo cp libstk.a /usr/local/lib/libstk.a

```

* https://github.com/sevagh/libmetro

Libmetro is a C++ library but it has a minimal C wrapper to work with cgo.

```

libmetro $ mkdir -p build && cd build && cmake .. -DBUILD_STATIC=ON && make && sudo make install

```

Finally, compile this project:

```

drum_machine $ go build .

```

cgo configuration is:

```

// #cgo LDFLAGS: /usr/local/lib/libmetro.a /usr/local/lib64/libsoundio.a /usr/local/lib/libstk.a -lstdc++ -lasound -lpulse -ljack -lm

// #include 

import "C"

```