Sound reproduction

As immersive instruments like the AlloSphere at UCSB evolve, the primary challenge shifts from simple visualization to the real-time synthesis of complex, multi-modal datasets. This paper explores the "Shaping Space as Information" framework through the lens of a Senior AI Architect. By implementing asynchronous data streams and generative spatial logic, we can transform immersive environments from passive display tools into active, forensic research engines where every pixel is a traceable data point. Traditional 3D rendering creates a significant computational bottleneck when handling massive datasets in a 360-degree environment. To scale information effectively, we propose an Asynchronous Data Stream model. • The Logic: Instead of rendering a static mesh, the system utilizes Python-driven "Spatial Culling." • The Application: The visual engine only processes high-resolution geometry within the user's immediate sensory field. This allows for the visualization of infinite datasets without latency, ensuring that the researcher's experience remains fluid and responsive to real-time inputs.

Supervised learning models have been applied to create good onset detection systems for musical audio signals. However, this always requires a large set of labeled training examples, and hand-labeling is quite tedious and time consuming. In this paper, we present a bootstrap learning approach to train an accurate note onset detection model. Audio alignment techniques are first used to find the correspondence between a symbolic music representation (such as MIDI data) and an acoustic recording. This alignment provides an initial estimate of note boundaries which can be used to train an onset detector. Once trained, the detector can be used to refine the initial set of note boundaries and training can be repeated. This iterative training process eliminates the need for hand-labeled audio. Tests show that this training method can improve an onset detector initially trained on synthetic data.

In the acoustic design of special spaces, such as studios, it may be required to realize diffuse reflection from a specific surface of a wall. Moreover, this requirement often changes with time or situation. In this research study, a new acoustic wall system, which has the potential to cope with such strict requirements, is proposed. The proposed system can vary the reflection characteristics by radiating additional sound from a source located behind the absorbing material. Therefore, if sound is not emitted from the source, it is possible to use the system as a mere sound absorbing material. This system can be used as a rigid wall, provided sound of equal amplitude to the incident sound is emitted from the source. Furthermore, a trial was carried out to make the characteristics of radiated sound being diffuse by using appropriate devices. Several models are proposed and the fundamental characteristics such as directivity were examined. The results show the possibilities for realizing uniform reflection directivity and variable absorption characteristics.

As immersive technologies such as Virtual Reality (VR) and Augmented Reality (AR) become increasingly prevalent, the demand for efficient, perceptually optimized audio compression has grown rapidly. Traditional audio codecs prioritize signal fidelity over perceptual relevance and often fail to preserve critical spatial cues under low-bitrate constraints. In this paper, we propose a novel perceptual neural audio codec tailored for low-bandwidth immersive applications, with an emphasis on preserving both auditory salience and spatial integrity. Our approach combines psychoacoustic feature extraction with a deep encoder-decoder architecture, conditioned on spatial metadata and trained using perceptual and spatially aware loss functions. The codec is evaluated across multiple datasets, including speech and ambient scenes, demonstrating superior performance over existing classical and neural codecs in terms of both objective metrics (e.g., SNR, LSD, ViSQO [1]) and subjective Mean Opinion Scores (MOS). Moreover, the lightweight model architecture enables real-time decoding on mobile hardware, making it suitable for deployment in VR/AR headsets, mobile devices, and low-power audio systems. This work establishes a foundation for scalable, high-quality spatial audio delivery under severe bandwidth constraints, advancing the state-of-the-art in perceptual audio coding for immersive media.

Pitch glide is an important effect that occurs in nearly all plucked string instruments. In essence, large amplitude waves traveling on a string during the note onset increases the string tension above its nominal value, and therefore cause the pitch to temporarily in- crease. Measurements are presented showing an exponential re- laxation of all the partial frequencies to their nominal values with a time-constant related to the decay rate of transverse waves prop- agating on the string. This exponential pitch trajectory is sup- ported by a simple physical model in which the increased tension is somewhat counterbalanced by the increased length of the string. Finally, a method for synthesizing the plucked string via a novel hybrid digital waveguide-modal synthesis model is presented with implementation details for time-varying resonators.

The production of high-quality educational video content faces significant bottlenecks due to manual recording requirements, technical expertise demands, and time-intensive editing processes. This paper presents an integrated AI-driven pipeline that automates avatar video generation by combining facial animation technology (SadTalker), neural text-to-speech synthesis (Google Cloud Neural2 TTS with SSML enhancement), and optimized video processing (FFmpeg). Our system architecture implements a three-phase workflow: (1) batch processing for scalable multi-avatar content generation, (2) interactive single-video creation with natural voice synthesis, and (3) quality enhancement through intelligent upscaling. Experimental results demonstrate a 90% reduction in production time (from 2+ hours to 15-25 minutes per 2-minute video) while maintaining 50-60% visual quality through full-frame preprocessing that preserves backgrounds and avatar context. The system achieves natural speech prosody through SSML markup integration and optimizes the quality-efficiency trade-off by generating at 256px resolution followed by 2-4× upscaling using Lanczos interpolation. Our batch processing implementation enables parallel generation of multiple avatar videos from structured scripts, reducing content creation barriers for educators without technical expertise. This research contributes a novel pipeline architecture for automated educational content production, demonstrating practical scalability and quality retention suitable for real-world deployment in online learning environments.

We show that the strongly spin-orbit coupled materials Bi 2 Te 3 and Sb 2 Te 3 and their derivatives belong to the Z 2 topological-insulator class. Using a combination of first-principles theoretical calculations and photoemission spectroscopy, we directly show that Bi 2 Te 3 is a large spin-orbit-induced indirect bulk band gap ( $ 150 meV) semiconductor whose surface is characterized by a single topological spin-Dirac cone. The electronic structure of self-doped Sb 2 Te 3 exhibits similar Z 2 topological properties. We demonstrate that the dynamics of spin-Dirac fermions can be controlled through systematic Mn doping, making these materials classes potentially suitable for topological device applications.

An ad-hoc Task Group within the SC-2-2 (Digital Input/Output Interfacing) Working Group is developing a more detailed specification for jitter in the embedded clock of the digital audio interface signal. The discussion and the intermediate conclusions of this work are presented in this paper as an invitation for informal comment and debate about the issues raised.

This study examines the temporal properties of a new form of auditory spatial plasticity called contextual plasticity. The contextual plasticity is a result of adaptation on the time scale of tens of seconds to minutes, but also shows changes at the scale of seconds. The study analyzes behavioral data from an experimental study in which horizontal-plane sound localization was examined in anechoic and reverberant environments (Kopčo et al., 2005). The results suggest that contextual plasticity has a bi-stable component that is strongly adaptive on the time scale of seconds, while overall the effect is building up over tens of seconds/minutes (Kopčo et al., 2007).

Directivity pattern of an ordinary loudspeaker becomes more directive at higher frequencies. However, because a single loudspeaker tends to radiate uniformly in all directions at low frequencies, reverberation from surrounding building walls may affect speech intelligibility when installing a multiple-loudspeaker system at crossroads. As an alternative, a sharply directive sound source is recommended to be used, but in many cases the directivity of an ordinary loudspeaker is less sharp at lower frequencies. Therefore, in order to overcome such a limitation, this paper discusses the possibility of using four loudspeakers under active control to realize a quadrupole radiation pattern in low frequency range. In this study, the radiation pattern of a primary loudspeaker and three secondary loudspeakers has been modelled. By placing the loudspeakers close together in the direction of 0 • , 90 • , 180 • , and 270 • , it was theoretically demonstrated that a quadrupole radiation pattern can be shaped in the target frequency range up to 600 Hz by simply controlling the directivity in three of four directions which are 45 • , 135 • , 225 • , and 315 • . Although, the radiation pattern model is far from realistic configurations and conditions, it is possible to realize a quadrupole radiation pattern in the low frequency range.

Public entertainment premises are significant sources of urban noise pollution. In this paper, indoor sound level spectra for different types of entertainment premises have been analyzed. While most of the spectra lie between two standardized curves proposed by the ISO 717-1:2013 (The International Organization for Standardization) standard, spectra recorded in discotheques show a pronounced spectral level below 160 Hz as a result of an intensive use of subwoofers. This indicates potentially inappropriate rating of sound insulation against such specific low-frequency noise by standardized spectrum adaptation terms. A modified spectrum adaptation term for a single-number rating of sound insulation under such circumstances has been proposed.

A novel closed-form method for designing fractional delay allpass filters is proposed. The design uses closed-form formulas and is based on truncating the coefficient vector of a Thiran allpass filter. While the resulting filters are non-optimal, they allow a wider approximation bandwidth than the Thiran allpass filter, which yields a maximally flat delay approximation at the zero frequency. Formulas have been derived to assist in choosing the two parameters, order and prototype order, for the new design. There is practically no upper limit for the filter order, since the method is not prone to numerical problems.

This paper presents an automated Public Address processing unit, using delay and magnitude response adjustment. The aim is to achieve a flat frequency response and delay adjustment between different physically-placed speakers at the measuring point, which is nowadays usually made manually by the sound technician. The adjustment is obtained using three signal processing operations to the audio signal: time delay adjustment, crossover filtering, and graphic equalization. The automation is in the calculation of different parameter sets: estimation of the time delay, the selection of a suitable crossover frequency, and calculation of the gains for a third-octave graphic equalizer. These automatic methods reduce time and effort in the calibration of line-array PA systems, since only three sine sweeps must be played through the sound system. Measurements have been conducted in an anechoic chamber using a 1:10 scale model of a line array system to verify the functioning of the automatic calibration and equalization methods.

This paper presents an automated Public Address processing unit, using delay and magnitude response adjustment. The aim is to achieve a flat frequency response and delay adjustment between different physically-placed speakers at the measuring point, which is nowadays usually made manually by the sound technician. The adjustment is obtained using three signal processing operations to the audio signal: time delay adjustment, crossover filtering, and graphic equalization. The automation is in the calculation of different parameter sets: estimation of the time delay, the selection of a suitable crossover frequency, and calculation of the gains for a third-octave graphic equalizer. These automatic methods reduce time and effort in the calibration of line-array PA systems, since only three sine sweeps must be played through the sound system. Measurements have been conducted in an anechoic chamber using a 1:10 scale model of a line array system to verify the functioning of the automatic ca...

This paper presents an audio-integrated motorcycle helmet that provides music, voice communication, and noise reduction functions. The effects of degrading the music and speech signals by the motorcycle noise were studied Experimental setup with real ambience of a person riding a motorcycle, wearing a helmet with active noise control (ANC), and listening to audio under noisy conditions was built. The performance of the audio-integrated system in reducing noise and preserving the quality of audio signals was evaluated using computer simulations and real-time experiments with recorded motorcycle noise.

Amethod for modeling room acoustics and auralizing the results in real time with amoving listener is presented. The acoustics modeling is based on the acoustic radiance transfer technique which is capable of modeling arbitrary reflection properties of different materials. The noveli dea of implementing this technique in frequency domain allows modeling of all frequencies at once as the time domain technique requires separate runs for each frequencyb and. Since the auralization of the results requires scaling, adding, and delaying responses as well as convolving them with head-related impulse responses, the massive parallel computation capacity of modern graphics hardware is utilized. Thus, realistic interactive walkthroughs are possible in typical room models with a stationary source.