songtools 0.1.0

Add your description here

songtools

A Python DSL for composing music from code. Notes carry expression, structure carries time. No magic strings, no duplicate logic or state.

Install

uv pip install -e .

Requires Python 3.14+ and sounddevice (PortAudio backend).

Core idea

Sound knows timbre, never time. Tune owns relative time. play / compile own absolute time (bpm). Combine them with operators and you get a tiny, composable music language.

Abstractions

Name	What it is
Sound	A voice (sample or synth). Knows timbre, never time.
Key	Harmonic context: root + scale + quality.
Tune	Anything arranged in time. Pure data — no methods that mutate.
Layer	Parallel voices. Shorter sides loop to LCM.
compile() / play()	The transport. Owns bpm and output.

Operators

Op	Meaning
Tune(a, b, c)	Slots of 1 beat; nesting subdivides
Layer(a, b, c)	Parallel voices; shorter sides loop to LCM
a + b	a, then b (sequence)
a * n	repeat
x @ y	realize: left plays right, in left's context

None is not used — use REST for a rest inside a Tune.

Vocabulary

• Notes: KeyRoot.C, KeyRoot.D, ... (chromatic, 0–11)
• Scales: Scale.MAJOR, Scale.MINOR, Scale.DORIAN
• Qualities: Quality.TRIAD, Quality.SEVENTH, Quality.NINTH, Quality.SUS2, Quality.SUS4, Quality.POWER
• Degrees: Degree.I, Degree.II, ... Degree.VII (also used as chord symbols)

Effects

Chain on any Sound with @:

Effect	What it does
Gain(amount)	Multiply amplitude
Decay(duration)	Exponential fade (timedelta)
LowPass(hertz)	One-pole lowpass filter
HighPass(hertz)	One-pole highpass filter
Echo(seconds)	Single-tap delay at −6 dB
Delay(seconds, fb, mix)	Feedback delay with dry/wet blend
Gate(seconds)	Hard cut to length
FadeIn(seconds)	Linear ramp up from silence
FadeOut(seconds)	Linear ramp down to silence
Reverse	Reverse the buffer
Drive(amount)	Tanh saturation
Clip(threshold)	Hard-limit peaks
Normalize(peak)	Scale so the loudest sample = peak
Tremolo(hertz, depth)	Amplitude modulation
BitCrush(bits)	Reduce bit depth (lo-fi)
Humanize(velocity)	Subtle gain variation

Quick start

1. Generate some sounds

python scripts/sounds.py     # writes pluck.wav, kick.wav, snare.wav, hat.wav

2. Compose

from datetime import timedelta

from songtools.files import WavFile
from songtools.keys import Key
from songtools.layers import Layer
from songtools.sounds import REST, Sound
from songtools.tunes import Tune
from songtools.types import (
    Decay,
    Degree,
    Echo,
    Gain,
    KeyRoot,
    LowPass,
    Pitch,
    Quality,
    Scale,
)

# --- Load or synthesize sounds ---
pluck = WavFile(Path("pluck.wav")).parse(Pitch(60))
kick  = WavFile(Path("kick.wav")).parse(Pitch(60))
hat   = WavFile(Path("hat.wav")).parse(Pitch(60))

# --- Shape them with effects ---
warm_pluck = pluck @ LowPass(1800) @ Decay(timedelta(milliseconds=900)) @ Gain(0.5)
hard_hat   = hat @ Gain(0.3)

# --- Tune to a key ---
key = Key(KeyRoot.C, Scale.MAJOR, Quality.TRIAD)
melody = Tune(
    warm_pluck @ key @ Degree.I,
    warm_pluck @ key @ Degree.IV,
    warm_pluck @ key @ Degree.V,
    warm_pluck @ key @ Degree.I,
)

# --- Arrange ---
beat = Tune(hard_hat, REST, hard_hat, REST) * 4
bass = Tune(kick, REST, kick, REST) * 4

song = Layer(melody, beat, bass)

# --- Render ---
song.compile(beats_per_minute=120).play()          # play live
WavFile(Path("out.wav")).save(song.compile(beats_per_minute=120))  # or save

See scripts/sounds.py and scripts/test.py for full working examples.

Rules of thumb

• Slot = 1 beat. Want it faster? Nest a Tune. Slower? Spread with REST. Global speed? bpm.
• Variations are data. Write them: groove * 3 + fill.
• Quality falls out of Key × degree. Never name a chord's quality — it emerges.
• Transpose a layer: change one Key constant. Drums don't move.
• @ chains left-to-right: sound @ key @ Degree.I, then add effects.

Project layout

src/songtools/
  types.py      # Buffer, Pitch, KeyRoot, Scale, Quality, Degree, effects
  sounds.py     # Sound, KeyedSound, REST
  keys.py       # Key (harmonic context)
  tunes.py      # Tune (sequential time)
  layers.py     # Layer (parallel time)
  transports.py # mixdown — the bpm-aware renderer
  files.py      # WavFile — load/save 16-bit mono WAV
scripts/
  sounds.py     # Synthesize pluck, kick, snare, hat
  test.py       # Full song example

License

MIT