AudioLabs - Real-Time Beat Tracking for Creative Music Production

Real-Time Beat Tracking for Creative Music Production

This is the accompanying website for the Bachelor Thesis [1].

Ole Frederik Müermann
Real-Time Beat Tracking for Creative Music Production
Bachelor Thesis, Friedrich-Alexander-Universität Erlangen-Nürnberg (FAU), 2024. Demo

@misc{Müermann24_RealTimeBeatMusicProduction_Bachelor_Thesis,
title       = {Real-Time Beat Tracking for Creative Music Production},
author      = {Ole Frederik Müermann},
year        = {2024},
note        = {Bachelor Thesis, Friedrich-Alexander-Universität Erlangen-Nürnberg (FAU)},
url-demo    = {https://audiolabs-erlangen.de/resources/MIR/2024-RealTimeBeat-MusicProduction}
}

**Figure 1.1.** The input audio signal undergoes beat tracking analysis to extract beat and tempo information. Control signals are generated, which automate a selected audio effect inside the DAW. The final output is the audio signal enhanced with effects synchronized with the detected beat.

Abstract

Music production and mixing involve the use of numerous audio effects, with parameters that need to be controlled. While manual control is possible, automation greatly enhances efficiency. This thesis presents a collection of creative digital audio effects and audio examples for real-time beat tracking. A new real-time procedure based on the predominant local pulse (PLP) method is introduced, as described in [2] and [3]. This beat tracker can extract two types of control signals: First, a low-frequency oscillation (LFO), and second, a confidence measure for beat stability. LFOs are used in audio signal processing where an oscillating signal with a frequency below the audible range (typically below 20 Hz) modulates various audio parameters, such as amplitude or filter frequency. The main focus of this thesis lies on a library of case studies on creative audio effects. Selected effects are presented using a demo track produced in the digital audio workstation (DAW) Ableton Live. The pre-generated control curves from the beat tracker are utilized to automate effect parameters synchronized with the beat of the music. Subsequently, the audio examples are demonstrated using Jupyter Notebooks. This study on creative music production aims to serve as a source of inspiration for music producers and sound engineers in potential applications and the further development of real-time beat tracking systems.

Audio Examples

Demo Track Overview

In this first example, we provide an overview of our demo track with (wet) and without (dry) effects. The figures show the waveform, song structure, and tempo (BPM) over time.

Table of Effects

With this table, we provide a description of the effects used for mixing the track. We include information on the control signals used, as well as the input signal to the beat tracker, the tempo range, the kernel size, and the lookahead parameter.

ID	Effect Description	Control Signals	(Input, Tempo Range, Kernel Size, Lookahead)
1	Arpeggiator	gamma-LFO	(Kick, 100-140, 6, 0)
2	Formant	alpha-LFO	(Drums, 120-170, 6, 0)
3	Tremolo - Vocal Stutter	gamma-LFO	(Kick, 400-560, 2, 30)
4	Reverb Ducking and Mixing	alpha-LFO, beta-confidence	(Kick, 100-140, 2, 0)
5	Kick and Bass Sidechaining	alpha-LFO, beta-confidence	(Kick, 100-140, 6, 0)
6	Wavetable Synthesis - Timbre	gamma-LFO	(Drums, 120-170, 6, 0)

Arpeggiator

An example of the arpeggiator effect...

**Figure 4.3.** Pipeline: The kick input signal undergoes beat tracking analysis to extract beat information. Control signals are generated. The γ-LFO is mapped to the arpeggiator rate of the pluck instrument. The final output is the effected pluck lead instrument, arpeggiated in a rhythm in sync with the beat.

Formant

An example of the formant effect...

**Figure 4.6.** Pipeline: The kick input signal undergoes beat tracking analysis to extract beat information. Control signals are generated. The α-LFO is mapped to the formant on the audio. The final output is the effected vocal, modified in sync with the beat.

Tremolo - Vocal Stutter

An example of the vocal stutter effect...

**Figure 4.10.** Pipeline: The kick input signal undergoes beat tracking analysis to extract beat information. Control signals are generated. The γ-LFO is mapped to the gain of the vocal. The final output is the vocal with a rhythmic stutter effect.

Reverb Ducking and Mixing

An example of reverb ducking and mixing...

**Figure 4.14.** Pipeline: The kick input signal undergoes beat tracking analysis. Control signals are generated. The α-LFO is mapped to duck the reverb on the piano. The β-confidence signal is used to mix in the reverb ducking only during the drop.

Kick and Bass Sidechaining

An example of kick and bass sidechaining...

**Figure 4.17.** Pipeline: The kick input signal undergoes beat tracking analysis to extract beat information. Control signals are generated and then shaped. These modified signals are mapped to the gain control of the bass. The final output is the effected bass, synchronized with the detected beat.

Wavetable Synthesis - Timbre

An example of wavetable synthesis timbre...

**Figure 4.21.** Pipeline: The drums input signal undergoes beat tracking analysis. Control signals are generated. The α-LFO is mapped to blend waveforms in the snythesizer. The final output is the effected bass synth.

References