Download Sound Matching with a Differentiable Karplus-Strong Algorithm
We present a self-supervised, event-based sound matching model using a differentiable extended Karplus-Strong algorithm. To avoid relying on external onset and fundamental frequency detectors, we explore training methodologies combining parameter losses on synthetic data with audio losses. We demonstrate that time-domain fractional delay interpolation provides gradient accuracy comparable to frequency-sampling while avoiding time-aliasing in highly resonant time-varying scenarios. Through systematic gradient analysis, we reveal that standard spectral losses provide no meaningful directional gradients for onset times, heavily degrading joint training. Training exclusively with parameter losses on synthetic data effectively learns fundamental frequency, timbral parameters, and onset times, but struggles to generalise to monophonic studio recordings of plucked guitar. External detectors combined with audio losses generalise best, isolating the model to timbre optimisation. While our Karplus-Strong decoder recovers interpretable parameters and naturally captures the transient characteristics of plucked guitar, Harmonics plus Noise baselines yield higher reconstruction fidelity by most metrics.
Download An Autonomous Method for Multi-Track Dynamic Range Compression
Dynamic range compression is a nonlinear audio effect that reduces the dynamic range of a signal and is frequently used as part of the process of mixing multi-track audio recordings. A system for automatically setting the parameters of multiple dynamic range compressors (one acting on each track of the multi-track mix) is described. The perceptual signal features loudness and loudness range are used to cross-adaptively control each compressor. The system is fully autonomous and includes six different modes of operation. These were compared and evaluated against a mix in which compressor settings were chosen by an expert audio mix engineer. Clear preferences were established for the different modes of operation, and it was found that the autonomous system was capable of producing audio mixes of approximately the same subjective quality as those produced by the expert engineer.
Download Automatic Mixing: Live Downmixing Stereo Panner
An automatic stereo panning algorithm intended for live multitrack downmixing has been researched. The algorithm uses spectral analysis to determine the panning position of sources. The method uses filter bank quantitative channel dependence, priority channel architecture and constrained rules to assign panning criteria. The algorithm attempts to minimize spectral masking by allocating similar spectra to different panning spaces. The algorithm has been implemented; results on its convergence, automatic panning space allocation, and left-right inter-channel phase relationship are presented.
Download Automatic Noise Gate Settings for Multitrack Drum Recordings
A method has been developed for automating the settings of a noise gate. The method has been applied to a kick drum track containing bleed from secondary drum sources and white noise. The optimal settings are found by maximising the signal to distortion ratio (SDR). The SDR has contributions from the distortion caused to the kick drum signal, and the residual bleed and noise. These two components are weighted, enabling the gate to be controlled by a single parameter. It is shown that the improvement in the SDR can be obtained when the two components of the SDR are approximated, enabling the optimal settings to be calculated from the noisy signal and a single kick drum hit. It is found that the optimal threshold is slightly above the peak level of the noise component of the signal.
Download Microphone Interference Reduction in Live Sound
When multiple microphones are used to reproduce multiple sources microphone interference, or bleed, can occur due to each microphone picking up more than one source. This paper proposes combining the crosstalk resistant adaptive noise canceller (CTRANC) algorithm with centred adaptive filters using an estimation of delay to suppress the interference, while making little change to the target signal. The proposed method is compared with similar methods in both the anechoic and echoic cases. The method is shown to outperform the other methods in the anechoic case while in the echoic case it is shown to perform less well at reducing the level of the interference but still introduces the least artefacts. Extension to the proposed method to the N source and microphone case is also discussed.
Download Differentiable All-Pole Filters for Time-Varying Audio Systems
Infinite impulse response filters are an essential building block of many time-varying audio systems, such as audio effects and synthesisers. However, their recursive structure impedes end-toend training of these systems using automatic differentiation. Although non-recursive filter approximations like frequency sampling and frame-based processing have been proposed and widely used in previous works, they cannot accurately reflect the gradient of the original system. We alleviate this difficulty by reexpressing a time-varying all-pole filter to backpropagate the gradients through itself, so the filter implementation is not bound to the technical limitations of automatic differentiation frameworks. This implementation can be employed within audio systems containing filters with poles for efficient gradient evaluation. We demonstrate its training efficiency and expressive capabilities for modelling real-world dynamic audio systems on a phaser, time-varying subtractive synthesiser, and feed-forward compressor. We make our code and audio samples available and provide the trained audio effect and synth models in a VST plugin1 .