Download Parametric Resynthesis of Measured Spatial Room Impulse Responses Spatial Room Impulse Responses (SRIRs) are fundamental to immersive audio rendering and have become a key focus of recent machine learning research in acoustics and auralization. Due to the high computational cost of direct convolution, spatial audio systems commonly employ artificial reverberation algorithms. However, these approaches often fail to accurately reproduce the spatial, temporal, and spectral characteristics of early reflections, leading to notable deviations from measured SRIRs. This paper presents a comprehensive framework for the analysis and efficient resynthesis of SRIRs captured with Spherical Microphone Arrays (SMAs). The proposed method accounts for hardware-induced artifacts, including scattering and spatial aliasing. Early reflections are reconstructed using a parametric approach based on the Herglotz analysis method, while late reverberation is synthesized using a Directional Feedback Delay Network (DFDN) with optimized filter-attenuation and correlation-matching. The proposed framework produces signals whose spatial correlation and Energy Decay Relief (EDR) closely match those of measured SRIRs, demonstrating its effectiveness for both real-time spatial audio rendering and realistic dataset generation for machine learning applications.
Download Compiling Differentiable Audio Graphs to Real-Time DSP Differentiable audio processors are habitually designed and optimised in machine-learning frameworks, but deploying them as real-time audio effects still often requires non-automatic implementation in a dedicated digital signal processing language. The translation is error-prone, demands an onerous verification process, and detaches research prototypes from usable production tools. That being so, we present ADAC, a compiler that lowers a trained model to a framework-agnostic intermediate representation and emits efficient FAUST code whose impulse response matches the source model to within floating-point arithmetic noise, direct paths included. The optimisation loop is made audible by replacing the model in a running plugin after each gradient step. The exported processor carries a small set of macro-controls that leave its stability intact. A stability certificate computed from the shipped parameters is checked before the plugin is built. At the demonstration, a feedback delay network is trained and exported to a working plugin.
Download Sound Effects Dataset Unification With the Universal Category System Sound effects (SFX) datasets and libraries often employ distinct tagging schemes, taxonomies, and metadata structures. This creates challenges for research on SFX classification and generation because incompatible taxonomies lead to siloed datasets that might require individualized approaches, result in non-comparable outcomes, and prevent data merging strategies. We propose a modular dataset relabeling framework that adopts the Universal Category System (UCS), an industry-standard hierarchical taxonomy for sound effects, as a shared structural foundation. This open-source framework enables us (i) to convert tags of existing datasets to UCS with a rule-based multi-stage pipeline and conflict resolution to achieve high automatic conversion rates, (ii) to suggest a stratified dataset split for the new labels, and (iii) to combine multiple datasets. To showcase the practical utility, we introduce the EnvSound-UCS dataset, a publicly available unified UCS-compliant dataset of environmental sounds with 58,057 sound clips from three sources: AudioSet, FSD50K, and ESC-50.
Download Multi-Source Extension and Hyperparameter Optimization of the DiffRIR Framework for Room Impulse Response Synthesis Efficient prediction of Room Impulse Responses (RIRs) is a cornerstone for immersive virtual acoustics and scalable room acoustic modeling. This study extends the DiffRIR framework – proposed by Wang et al. in Hearing Anything Anywhere – by introducing a multi-source training logic and systematically optimizing its convergence behavior to overcome the inherent limitations of the original framework. Our results reveal that multi-source training acts as implicit data augmentation, where the resulting increase in spatial entropy enhances the model's spectral accuracy. Furthermore, we demonstrate that the model exhibits remarkable robustness against geometric inaccuracies, maintaining numerical stability even with source positional offsets of up to 4 m in single-source baseline evaluations. By identifying a learning rate of 3×10⁻², we were able to reduce the training duration to 23% of the original baseline without compromising prediction accuracy. While the increased complexity of multi-source fields necessitates a trade-off in temporal precision – quantified via our newly integrated Energy Decay Convergence (EDC) metric – this research provides an efficient and resilient solution for acoustic simulations in complex environments.
Download Fourier Neural Operators for Sample-Rate-Independent Virtual Analog Modeling Neural networks that operate directly on time-domain signals are widely used for virtual analog (VA) modeling. A key limitation of these models is their dependence on the sampling rate used during training, which becomes implicitly encoded in the learned parameters, so that changing it generally alters the realized dynamics. Although architectural modifications to recurrent neural networks have been proposed to enable sample-rate independent operation, these approaches are inherently tailored to upsampling and do not accommodate downsampling scenarios. In this manuscript, we present a VA modeling framework based on Fourier Neural Operators (FNOs) adapted to process fixed-duration audio frames. The proposed formulation defines the learned mapping over a fixed temporal support and evaluates it on uniform grids of different densities, so that a model trained at a single sampling rate can be applied at unseen sampling resolutions. Numerical results on a nonlinear transistor circuit show that the proposed model achieves competitive accuracy in upsampling scenarios while remaining directly applicable to downsampling, unlike a sample-rate independent baseline recurrent architecture.
Download Probing Low-Level Acoustic Attribute Encoding in CLAP Audio Embeddings This work analyzes CLAP audio embeddings through a probing framework, studying the encoding of reverberation (RT60), loudness (LUFS), spectral content (SC), and relative pitch (RP). Results show that all attributes are reliably recoverable from CLAP embeddings, with RT60, LUFS, and RP approximately linearly encoded, while SC requires non-linear probes. The identified patterns generalize across eight additional audio foundation models.
Download Parameter Estimation via Differentiable Modal Plate Synthesis We present our submission to Task A of the 1st DAFx Parameter Estimation Challenge, which concerns the estimation of the physical parameters of a vibrating plate from a synthetic impulse response. Our approach introduces a differentiable modal plate synthesizer and estimates the plate parameters through inference-time gradient-based optimization of the synthesizer parameters. The six target parameters are recovered by minimizing a multi-scale spectral loss via backpropagation through the differentiable plate model. To handle the non-convexity of the loss landscape, we adopt a two-phase training strategy consisting of multiple short-term probe optimizations, followed by full-scale refinement initialized from the best candidate. We evaluate the approach on eight impulse responses synthesized with the official challenge dataset generator. Compared with a constant-value predictor and the particle swarm optimization baseline provided by the challenge, the proposed method reduces the prediction error by approximately one order of magnitude.
Download Enhancing Automatic Chord Recognition via Pseudo-Labeling and Knowledge Distillation Automatic Chord Recognition (ACR) is constrained by the scarcity of aligned chord annotations, which are costly to acquire. At the same time, open-weight pre-trained models are more accessible than their proprietary training data. In this work, we present a two-stage training pipeline that leverages pre-trained models together with unlabeled audio. The proposed method decouples training into two stages. In the first stage, we use the pre-trained BTC model as a teacher to generate pseudo-labels for over 1,000 hours of diverse unlabeled audio and train a student model solely on these pseudo-labels. In the second stage, the student is continually trained on ground-truth labels as they become available. To prevent catastrophic forgetting of the representations learned in the first stage, we apply selective knowledge distillation (KD) from the teacher as a regularizer. In our experiments, two models (BTC, 2E1D) were used as students. In Stage 1, using only pseudo-labels, the BTC student achieves about 99% of the teacher's performance, while the 2E1D model achieves about 97% of the teacher's performance across seven standard mir_eval metrics. After continual training with labeled data in Stage 2, the resulting BTC student model consistently surpasses both the traditional supervised learning baseline and the original pre-trained teacher model across all metrics. The resulting 2E1D student model also outperforms the supervised baseline and approaches teacher-level performance, with both models demonstrating substantial gains on rare chord qualities.
Download Peak-Residual Modal Estimation with Learned Calibration and High-Band Density Correction ★ This paper describes two related submissions to Task B of the 1st DAFx Parameter Estimation Challenge. Both estimate modal frequency, decay, and gain directly from an unnormalised plate impulse response without using plate parameters, the excluded analytical modal-frequency law, or official-test ground truth. The primary system constructs a large candidate pool through prominence-graded spectral peak picking, iterative residual analysis, multi-view consensus, band-wise budgeting, and a learned file-level mode-count target. Raw decay and gain estimates are then corrected by a small mode-wise neural network that is not allowed to move frequencies or change the number of rows. A secondary variant addresses suspected high-frequency under-counting with a separately gated, non-oracle density-fill stage in the 6–10 kHz band. The paper reports development diagnostics, reproducibility information, and descriptive statistics for the 16 official outputs. The two variants expose a deliberate precision–recall trade-off: one preserves a visible spectral justification for every row, while the other tests bounded hidden-multiplicity augmentation in densely overlapped regions.
Download A Multi-Resolution Spectrogram Approach for Estimating the Physical Parameters of a Plate Reverb The ResNet-18 image classification model is employed to determine the physical parameters of a plate reverb from a recording of the impulse response. The model is adapted to derive parameters using normalized and down-sampled multi-resolution spectrograms computed from the provided impulse responses (IRs). To refine the prediction of the output location, the spectral phase response is also included as an additional input channel to the network since multiple output locations can give the same magnitude response for high-order resonant modes. On a 5000 IR validation set, our model achieves an average normalized mean squared error (NMSE) of 0.02920 across all parameters, with the lowest average NMSE occurring for parameters yo (0.00228), Ly (0.00347), and xo (0.00574).