Block based synthesis and music library for Python. Klang is German for sound.

We use Python bindings for PortAudio and RtMidi. On Mac they can be installed via Homebrew.

Klang can be installed via PyPi / pip or directly via setup.py. Note that there are some audio C extensions. Python fallbacks exists.

python3 setup.py build_ext --inplace

For developing you can link your working copy with

python3 setup.py develop

Tests can be run via with

python3 setup.py test

As always when programming with sound: Unplug your headphones or be very sure of what you are doing! Also with low headphone volume bugs in the code can result in very unpleasant loud noises which could probably impair your hearing. Be careful!

Klang provides various audio related blocks. Every block can have multiple in-and outputs and by connecting them with each other we can define our network. Once we are finished with patching we can run our network with by calling run_klang(*blocks). This function only needs some blocks which belong to the network. It will then automatically discovers the other blocks of the network and deduce an appropriate block execution order.

In the following script we create a 440 Hz sine oscillator which output gets send to the sound card.

from klang.audio import Oscillator, Dac
from klang.klang import run_klang

# Init blocks
osc = Oscillator(frequency=440.)
dac = Dac(nChannels=1)

# Define network
osc.output.connect(dac.input)

# Run it
run_klang(dac)

Audio can be written to disk as a WAV file with the filepath argument.

run_klang(*blocks, filepath='some/filepath.wav')

There are two different types of connections in Klang:

Value

These connection can hold any kind of Python object which will be propagated through the network and updated during each cycle. Most commonly these are numpy arrays holding audio or modulation signals (Input and Output classes).

Message

Discrete messages. Each message input has its own internal message queue. Most commonly Note messages (MessageInput and MessageOutput classes).

There are also corresponding Relay connections (Relay and MessageRelay classes). These are used to build composite blocks (blocks which contain there own network of child blocks). Relays can be used to interface between the inside and outside of an composite block.

The connect method can be used to connect inputs and outputs with each other. Note that it is always possible to connect one output to multiple inputs but not the other way round. As a shorthand there are two overloaded operators:

• Pipe operator |: Connect multiple blocks in series.
• Mix operator +: Mix multiple value outputs together.

# Pipe operator
a | b | c

# Is equivalanet to:
# >>> a.output.connect(b.input)
# ... b.output.connect(c.input)

# Mix operator
mixer = a + b + c

# Is equivalanet to:
# >>> mixer = Mixer(nInputs=0)
# ... mixer.add_new_channel()
# ... a.output.connect(mixer.inputs[-1])
# ... mixer.add_new_channel()
# ... b.output.connect(mixer.inputs[-1])
# ... mixer.add_new_channel()
# ... c.output.connect(mixer.inputs[-1])

See the examples directory with a couple example script which illustrate the core functionality of Klang. Currently there are:

• hello_world.py: 440 Hz sine wave generator
• aeolian_arp.py: More fun with random ever changing arpeggios.
• arpeggiator_demo.py: Two synthesizer patch with an arpeggiator and some sound effects
• audio_file_demo.py: Looped audio file playback (gong.wav sample) with audio effects
• haunting_envelopes.py: Multiple oscillators controlled by looping envelopes
• micro_rhythm_demo.py: Kick and Hi-Hat pattern where the latter is phrased with a micro rhythm
• reverberation_demo.py: Ambient loop showcasing the reverb effect.
• sequencer_demo.py: Techno patch with sequencer
• synthesizer_demo.py: This has to be started as root. Computer keyboard playable monophonic synthesizer
• tempo_aware_effects.py: Modulated noise with time synced effects

PEP8 / Google flavored. With the one exception for variable and argument names (camelCase). Function and in methods are snake_case().

• Alexander Theler (GitHub)

Thanks for the support and inputs!

• Nico Neureiter
• Andreas Steiner
• Lawrence Markwalder