Paper Archive

This paper addresses the task of lyrics-to-audio alignment, which involves synchronizing textual lyrics with corresponding music audio. Most publicly available datasets for this task provide annotations only at the line or word level. This poses a challenge for training lyrics-to-audio models due to the lack of frame-wise phoneme labels. However, we find that phoneme labels can be partially derived from word-level annotations: for single-phoneme words, all frames corresponding to the word can be labeled with the same phoneme; for multi-phoneme words, phoneme labels can be assigned at the first and last frames of the word. To leverage this partial information, we construct a mask for those frames and propose a masked frame-wise cross-entropy (CE) loss that considers only frames with known phoneme labels. As a baseline model, we adopt an autoencoder trained with a Connectionist Temporal Classification (CTC) loss and a reconstruction loss. We then enhance the training process by incorporating the proposed framewise masked CE loss. Experimental results show that incorporating the frame-wise masked CE loss improves alignment performance. In comparison to other state-of-the art models, our model provides a comparable Mean Absolute Error (MAE) of 0.216 seconds and a top Median Absolute Error (MedAE) of 0.041 seconds on the testing Jamendo dataset.

Download

Decorrelation filters transform mono audio into multiple decorrelated copies. This paper introduces a novel decorrelation filter design based on a resonator bank, which produces a sum of over a thousand exponentially decaying sinusoids. A headphone listening test was used to identify the minimum inter-channel time delays that perceptually match ERB-filtered coherent noise to corresponding incoherent noise. The decay rate of each resonator is set based on a group delay profile determined by the listening test results at its corresponding frequency. Furthermore, the delays from the test are used to refine frequency-dependent windowing in coherence estimation, which we argue represents the perceptually most accurate way of assessing interaural coherence. This coherence measure then guides an optimization process that adjusts the initial phases of the sinusoids to minimize the coherence between two instances of the resonator-based decorrelator. The delay results establish the necessary group delay per ERB for effective decorrelation, revealing higher-than-expected values, particularly at higher frequencies. For comparison, the optimization is also performed using two previously proposed group-delay profiles: one based on the period of the ERB band center frequency and another based on the maximum group-delay limit before introducing smearing. The results indicate that the perceptually informed profile achieves equal decorrelation to the latter profile while smearing less at high frequencies. Overall, optimizing the phase response of the proposed decorrelator yields significantly lower coherence compared to using a random phase.

Download

Combining different room acoustic modelling methods could provide a better balance between perceptual plausibility and computational efficiency than using a single and potentially more computationally expensive model. In this work, a hybrid acoustic modelling system that integrates ray tracing (RT) with an advanced feedback delay network (FDN) is designed to generate perceptually plausible RIRs. A multiple stimuli with hidden reference and anchor (MUSHRA) test and a two-alternative-forced-choice (2AFC) discrimination task have been conducted to compare the proposed method against ground truth recordings and conventional RT-based approaches. The results show that the proposed system delivers robust performance in various scenarios, achieving highly plausible reverberation synthesis.

Download

Given the speedy computation of the FFT in current computer hardware, there are new possibilities for examining transformations for very long sounds. A zero-phase version of any audio signal can be obtained by zeroing the phase angle of its complex spectrum and taking the inverse FFT. This paper recommends additional processing steps, including zero-padding, transient suppression at the signal’s start and end, and gain compensation, to enhance the resulting sound quality. As a result, a sound with the same spectral characteristics as the original one, but with different temporal events, is obtained. Repeating rhythm patterns are retained, however. Zero-phase sounds are palindromic in the sense that they are symmetric in time. A comparison of the zero-phase conversion to the autocorrelation function helps to understand its properties, such as why the rhythm of the original sound is emphasized. It is also argued that the zero-phase signal has the same autocorrelation function as the original sound. One exciting variation of the method is to apply the method separately to the real and imaginary parts of the spectrum to produce a stereo effect. A frame-based technique enables the use of the zero-phase conversion in real-time audio processing. The zero-phase conversion is another member of the giant FFT toolset, allowing the modification of sampled sounds, such as drum loops or entire songs.

Download

This study evaluates the performance of five objective audio quality metrics—PEAQ Basic, PEAQ Advanced, PEMO-Q, ViSQOL, and HAAQI —in the context of digital music production. Unlike previous comparisons, we focus on their suitability for production environments, an area currently underexplored in existing research. Twelve audio examples were tested using two evaluation types: an effectiveness test under progressively increasing degradations (hum, hiss, clipping, glitches) and a robustness test under fixed-level, randomly fluctuating degradations. In the effectiveness test, HAAQI, PEMO-Q, and PEAQ Basic effectively tracked degradation changes, while PEAQ Advanced failed consistently and ViSQOL showed low sensitivity to hum and glitches. In the robustness test, ViSQOL and HAAQI demonstrated the highest consistency, with average standard deviations of 0.004 and 0.007, respectively, followed by PEMO-Q (0.021), PEAQ Basic (0.057), and PEAQ Advanced (0.065). However, ViSQOL also showed low variability across audio examples, suggesting limited genre sensitivity. These findings highlight the strengths and limitations of each metric for music production, specifically quality measurement with compressed audio. The source code and dataset will be made publicly available upon publication.

Download

The acoustic piano and its sound production mechanisms have been extensively studied in the field of acoustics. Similarly, digital piano synthesis has been the focus of numerous signal processing research studies. However, the role of the piano action in shaping the dynamics and nuances of piano sound has received less attention, particularly in the context of digital pianos. Digital pianos are well-established commercial instruments that typically use weighted keys with two or three sensors to measure the average key velocity—this being the only input to a sampling synthesis engine. In this study, we investigate whether this simplified measurement method adequately captures the full dynamic behavior of the original piano action. After a brief review of the state of the art, we describe an experimental setup designed to measure physical properties of the keys and hammers of a piano. This setup enables high-precision readings of acceleration, velocity, and position for both the key and hammer across various dynamic levels. Through extensive data analysis, we examine their relationships and identify the optimal key position for velocity measurement. We also analyze a digital piano key to determine where the average key velocity is measured and compare it with our proposed optimal timing. We find that the instantaneous key velocity just before let-off correlates most strongly with hammer impact velocity, indicating a target for improved sensing; however, due to the limitations of discrete velocity sensing this optimization alone may not suffice to replicate the nuanced expressiveness of acoustic piano touch. This study represents the first step in a broader research effort aimed at linking piano touch, dynamics, and sound production.

Download

This paper introduces a digital reconstruction of the morphophone, a complex magnetophonic device developed in the 1950s within the laboratories of the GRM (Groupe de Recherches Musicales) in Paris. The analysis, design, and implementation methodologies underlying the Digital Morphophone Environment are discussed. Based on a detailed review of historical sources and limited documentation – including a small body of literature and, most notably, archival images – the core operational principles of the morphophone have been modeled within the MAX visual programming environment. The main goals of this work are, on the one hand, to study and make accessible a now obsolete and unavailable tool, and on the other, to provide the opportunity for new explorations in computer music and research.

Download

Estimations of panning attributes are an important feature to extract from a piece of recorded music, with downstream uses such as classification, quality assessment, and listening enhancement. While several algorithms exist in the literature, there is currently no comparison between them and no studies to suggest which one is most suitable for any particular task. This paper compares four algorithms for extracting amplitude panning features with respect to their suitability for unsupervised learning. It finds synchronicities between them and analyses their results on a small set of commercial music excerpts chosen for their distinct panning features. The ability of each algorithm to differentiate between the tracks is analysed. The results can be used in future work to either select the most appropriate panning feature algorithm or create a version customized for a particular task.

Download

Most real systems, from acoustics to analog electronics, are characterised by bidirectional coupling amongst elements rather than neat, unidirectional signal flows between self-contained modules. Integrating digital processing into physical domains becomes a significant engineering challenge when the application requires bidirectional coupling across the physical-digital boundary rather than separate, well-defined inputs and outputs. We introduce an approach to hybrid analog-digital audio processing using synthetic impedance: digitally simulated circuit elements integrated into an otherwise analog circuit. This approach combines the physicality and classic character of analog audio circuits alongside the precision and flexibility of digital signal processing (DSP). Our impedance synthesis system consists of a voltage-controlled current source and a microcontroller-based DSP system. We demonstrate our technique through modifying an iconic guitar distortion pedal, the Boss DS-1, showing the ability of the synthetic impedance to both replicate and extend the behaviour of the pedal’s diode clipping stage. We discuss the behaviour of the synthetic impedance in isolated laboratory conditions and in the DS-1 pedal, highlighting the technical and creative potential of the technique as well as its practical limitations and future extensions.

Download

The common-slope model is used to model late reverberation of complex room geometries such as multiple coupled rooms. The model fits band-limited room impulse responses using a set of common decay rates, with amplitudes varying based on listener positions. This paper investigates amplitude estimation methods within the common-slope model framework. We compare several traditional least squares estimation methods and propose using LINEX regression, a Maximum Likelihood approach using logsquared RIR statistics. Through statistical analysis and simulation tests, we demonstrate that LINEX regression improves accuracy and reduces bias when compared to traditional methods.

Download