Additive & Spectral: Partial Tracking

Partial tracking is the process of detecting single sinusoidal components in a signal and obtaining their individual amplitude, freuquency and phase trajectories.


  • STFT

    A short term fourier transform segments a signal into frames of equal length.

  • Fundamental Frequency Estimation - YIN (de Cheveigné et al, 2002) - Swipe (Camacho, 2007)

  • Peak Detection

    For every STFT frame, local maxima are calculated in the range of integer multiples of the fundamental frequency.


Trajectories of partial amplitudes for a violin sound.


Trajectories of partial frequencies for a violin sound.


Trajectories of unwrapped partial phases for a violin sound.

Wait for (Audio) Hardware in systemd

Wait for (Audio) Hardware

In some cases it can be necessary to wait for a specific device, like a USB interface, before starting a service with systemd. This is the case for the JACK server in some setups.

With system.device

''systemd.device'' makes it possible to let services depend on the state of specific hardware. The following instructions work only for a static hardware setup and can fail if the hardware configuration is changed.

First: Find if systemd has a unit configuration file for the device. The following code lists all devices:

$ systemctl --all --full -t device

This will be a lot of output. Narrow down the search by grepping for the ALSA name of the device - in this case 'Track':

$ systemctl --all --full -t device | grep Track

The output looks as follows in this case:

sys-devices-pci0000:00-0000:00:14.0-usb2-2\x2d2-2\x2d2:1.0-sound-card1.device                               loaded active plugged M-Audio Fast Track

The '.device' identifier can now be used in the UNIT description's parts After and Requires:

Description=Jack audio server sys-devices-pci0000:00-0000:00:14.0-usb2-2\x2d2-2\x2d2:1.0-sound-card1.device

Waveguide Strings in Faust

Waveguides are physical models of one-dimensional oscillators. They can be used for emulating strings, reeds and other components. A more detailed explanation is featured in the Sound Synthesis Introduction. The following example implements a string with losses, excited with a triangular function.


Faust Code

Load this example in the Faust online IDE for a quick start.


// use '(pm.)l2s' to calculate number of samples
// from length in meters:

segment(maxLength,length) = waveguide(nMax,n)
nMax = maxLength : l2s;
n = length : l2s/2;

// one lowpass terminator
fc = hslider("lowpass",1000,10,10000,1);
rt = rTermination(basicBlock,*(-1) : si.smooth(1.0-2*(fc/ma.SR)));

// one gain terminator with control
gain = hslider("gain",0.99,0,1,0.01);
lt = lTermination(*(-1)* gain,basicBlock);

idString(length,pos,excite) = endChain(wg)

nUp   = length*pos;

nDown = length*(1-pos);

wg = chain(lt : segment(6,nUp) : in(excite) : out : segment(6,nDown) : rt); // waveguide chain

length = hslider("length",1,0.1,10,0.01);
process = idString(length,0.15, button("pluck")) <: _,_;



  • Romain Michon, Julius Smith, Chris Chafe, Ge Wang, and Matthew Wright. The faust physical modeling library: a modular playground for the digital luthier. In International Faust Conference. 2018.
    [details] [BibTeX▼]


Binaural Spatialization with SC-HOA

The SC-HOA library by Florian Grond is a feature-rich toolbox for working with Ambisonics and binaural synthesis in SuperCollider. Once installed, it is well documented inside the SC help files. Additional information and install instructions are part of the Git repository. This section gives a brief introduction into the solution used for the SPRAWL server.

Installing SC-HOA

The SC-HOA library is shipped as a so called Quark and it can be installed from inside SC. Besides a GUI-based way, a single command is enough to install the complete library with all objects and classes in the system's directories:


Make sure to reboot the interpreter after installing the Quark. The external classes need to be compiled.

To find out where SC has installed your external, run:


Encoding a 1st Order Source

The Ambisonics Bus

First thing to create is an audio rate bus for the encoded Ambisonics signal. The bus size depends on the Ambisonics order M, following the formula \(N = (M+1)^2\). For simplicity, this example uses first order:


// create the Ambisonics mix bus

~order     = 1;
~nHOA      = (pow(~order+1,2)).asInteger;
~ambi_BUS  =,~nHOA);

The channels of this bus correspond to the spherical harmonics. They encode the overall pressure and the distribution along the three basic dimensions:


The Encoder

The SC-HOA library includes different encoders. This example uses the HOASphericalHarmonics class. This simple encoder can set the angles of incidence (azimuth, elevation) in spherical coordinates. Angles are controlled in radians:

  • azimuth=0 with elevation=0 is a signal straight ahead

  • azimuth=-pi/2 is hard left

  • azimuth=pi/2 is hard right

  • azimuth=pi is in the back.

  • elevation=pi/2 is on the top

  • elevation=-pi/2 is on the bottom

This example uses a sawtooth signal as mono input and calculates the four Ambisonics channels.

~encoder_A = {arg azim=0, elev=0;,HOASphericalHarmonics.coefN3D(~order,azim,elev)*;

The Ambisonics bus can be monitored and the angles of the source can be set, manually:


// set parameters


Change the angles of the encoder and check whether the Ambisonics buses behave as expected. (Use multiples of pi/2.)

The Decoder

The SC-HOA library features default binaural impulse responses, which need to be loaded first:

// load binaural IRs for the decoder

Afterwards, a first order HOABinaural decoder is fed with the encoded Ambisonics signal. It needs to be placed after the encoder node to get an audible output to the left and right channels. This output is the actual binaural signal for headphone use.

~decoder = {,,~nHOA))}.play;


Listen to the decoded signal and change the angles.

Panning Multiple Sources

Working with multiple sources requires a dedicated encoder for each source. All encoded signals are subsequently routed to the same Ambisonics bus and a single decoder is used to create the binaural signal. The angles of all sources can be set, individually.

~encoder_B = {arg azim=0, elev=0;,HOASphericalHarmonics.coefN3D(~order,azim,elev)*}.play;



Exercise I

Use the mouse for a continuous control of a source's angles.

Exercise II

Add a control for the source distance to the encoder.

Exercise III

Increase the Ambisonics order and compare the results.

Exercise IV

Use OSC messages to control the positions of multiple sources.

Audio Buffers

Most systems for digital signal processing and music programming process audio in chunks, which are defined by a so called buffer size. These buffer sizes are usually powers of 2, usually ranging from $16$ samples - which can be considered a small buffer size - to $2048$ samples (and more). Most applications, like DAWs and hardware interfaces allow the user to select this parameter. Technically this means that a system collects (or buffers) single samples - for example from an ADC (analog-digital-converter) - until the buffer is filled. This compensates irregularities in the speed of execution for single operations and ensures a jitter-free processing.


The choice of the buffer size $N$ is usually a trade-off between processor load and system latency. Small buffers require faster processing whereas large buffers keep the user waiting until a buffer has been filled. In combination with the sampling rate $f_s$, the buffer-dependent latency can be calculated as follows:

$$ \tau = \frac{N}{f_s} $$

Round trip latency usually considers both the input and output buffers, thus doubling the latency. For a system running at $48\ \mathrm{kHz}$ with a buffer size of $128$ samples - a typical size for a decent prosumer setup - this results in a round trip latency of $5.5\ \mathrm{ms}$. This value is low enough to allow a perceptually satisfying interaction with the system. When exceeding the $10\ \mathrm{ms}$ threshold it is likely that percussions and other timing-critical instruments experience disrupting latency.

Buffers in Programming

In higher level programming environments like PD, MAX, SuperCollider or Faust (depending on the way it is used), users usually do not need to deal with the buffer size. When programming in C or C++, most frameworks and APIs offer a processing routine which is based on the buffer size. This accounts for solutions like JUCE or the JACK API, but also when programming externals or extensions for the above mentioned higher level environments. These processing routines, also referred to as callback, are called by an interrupt once the the hardware is ready to process the next buffer.

Working with Images

Getting the sprawl access point image

Download the operating system image for the sprawl access points! According to its postfix .xz the image is compressed with the XZ compression algorithm. Before writing the image to the sd card it has to be decompressed.

On Ubuntu/Debian you can install xz-utils and use unxz:

$ sudo apt-get install xz-utils
unxz sprawl_pi_image_20200628_shrinked.img.xz

On MacOS try the unarchiver. On Windows try the usual way to decompress files or download xz-utils.

Writing images on sd card


➜  ~ diskutil list
/dev/disk0 (internal, physical):
   #:                       TYPE NAME                    SIZE       IDENTIFIER
   0:      GUID_partition_scheme                        *1.0 TB     disk0
   1:                        EFI EFI                     209.7 MB   disk0s1
   2:                 Apple_APFS Container disk1         1000.0 GB  disk0s2

/dev/disk1 (synthesized):
   #:                       TYPE NAME                    SIZE       IDENTIFIER
   0:      APFS Container Scheme -                      +1000.0 GB  disk1
                                 Physical Store disk0s2
   1:                APFS Volume OWC Aura Pro SSD - Data 520.8 GB   disk1s1
   2:                APFS Volume Preboot                 82.6 MB    disk1s2
   3:                APFS Volume Recovery                525.8 MB   disk1s3
   4:                APFS Volume VM                      2.1 GB     disk1s4
   5:                APFS Volume OWC Aura Pro SSD        11.2 GB    disk1s5

/dev/disk2 (internal, physical):
   #:                       TYPE NAME                    SIZE       IDENTIFIER
   0:     FDisk_partition_scheme                        *63.9 GB    disk2
   1:             Windows_FAT_32 boot                    268.4 MB   disk2s1
   2:                      Linux                         63.6 GB    disk2s2

disk2 seems to be the sd card with 64 GB. To access the raw device /dev/rdisk2 is used. To write to this device you have to be root (administrator). Unmount the device before using dd.

diskutil unmountDisk /dev/disk2
sudo dd if=rpi.img of=/dev/rdisk2 bs=4M


$ sudo dd if=rpi.img of=/dev/disk/by-id/
$ sudo dd if=rpi.img of=/dev/disk/by-id/mmc-SDC_0x000000e2 bs=4M status=progress


On Windows the easiest way to write an image to a sd card is to use a dedicated application like Balena Etcher.

Wavefolding Example

The following example calculates the spectrum of a sinusoidal function, folded with a sinusoidal transfer function:



Time Domain:

Frequency Domain:

Envelopes: ADSR

Envelopes are an essential part of control in electronic music and computer music. They are used to shape the characteristics of sound or other processes over time and are an integral part of synthesizers. Since they are that basic and versatile, they will be introduced in this early section.

One of the most common envelopes, already featured in early synthesizers and in prominent examples as the MiniMoog, is the ADSR envelope (Hiyoshi, 1979). It is comprised of four segments:

  • Attack

  • Decay

  • Sustain

  • Release

Attack time, decay time and release time can usually be controlled by the user via dials or sliders, whereas the sustain time depends on the duration a key is pressed and the sustain level may depend on the stroke velocity. Depending on the settings, the ADSR model can generate amplitude and timbral envelopes for slowly evolving sounds like strings or sounds with sharp attacks and release:

Your browser does not support the HTML5 canvas tag

Attack Time:

Decay Time:

Sustain Level:

Sustain Time:

Release Time:

When used in synthesizers, this envelope can be used to control the overall level or the timbre - for example through the cutoff frequency of a filter or by means of partial amplitudes.



  • Teruo Hiyoshi, Akira Nakada, Tsutomu Suzuki, Eiichiro Aoki, and Eiichi Yamaga. Envelope generator. December 18 1979. US Patent 4,178,826.
    [details] [BibTeX▼]

Working with Groups

Creating Groups

Groups - or group nodes - can be a very useful concept for keeping track of the signal flow. Without any further actions, all nodes are placed in the default Group 1. Additional groups can be arranged regarding the order of execution. A new group can be added from sclang as follows:

~g1 =;

Relative Group Positions

As with nodes, further groups can be added in relation to existing groups. The following action makes sure that a new group will be placed after the previously defined group:

~g2 = Group.after(~g1);

Nested Groups

Groups can contain other groups, allowing a hierarchical structure of nodes:

~g3 = Group.head(~g2);

More on Groups

The group object allows many more actions. They are listed in the SC documentation on groups. After adding another group before the third one

~g4 = Group.before(~g3);

the server node structure looks as follows:


The server does not know the groups by their variable names in sclang. Hence they are numerated. Node indices - or IDs - of groups can be assessed from the language:




Use groups to sort the nodes from the example in the section on Combining Nodes