In this paper
AbstractIntroductionSignal Model and PreliminariesSignal ModelProblem FormulationPreliminariesLinear EncoderDiffusion ModelsDiffusion Posterior SamplingADEUPSDPS Formulation for Ambisonics ReconstructionInferenceOrder-Masking Prior TrainingDiscussionEmpirical EvaluationExperimental SetupTraining DataEvaluation DataMethodologyBenchmarksMethodologyEvaluation MetricsResultsFull- and Semi-Agnostic Methods EvaluationArray Specific Methods EvaluationAblation StudyListening ExperimentSetupMethodologyResultsConclusionsReferencesWorksEngineering
Array-Agnostic High-Order Ambisonics Encoding via Diffusion Posterior Sampling
Abstract
Ambisonics provides a device-independent representation of three-dimensional sound fields; however, extracting these coefficients from practical microphone array recordings remains strongly dependent on array geometry, acoustic transfer functions, and spatial sampling resolution. This pipeline is typically divided into two distinct tasks: encoding, which entails reconstructing the physically resolvable spatial components, and upscaling, which requires generating higher-order coefficients. Model-based approaches often fail on irregular arrays due to ill-conditioned transfer matrices, whereas conventional neural encoders remain strictly tied to the specific acquisition setup seen during training. Furthermore, existing upscaling methods usually assume ideal low-order inputs, ignoring device-specific array artifacts. We present ADEUPS, a unified framework that couples an array-independent Ambisonic diffusion prior with an explicit physical acquisition model during inference, enabling zero-shot encoding and upscaling across diverse array configurations. We train our prior using an order-masking strategy and perform inference via score blending, which we demonstrate to be crucial for solving the underlying inverse problem. Through extensive numerical evaluations across various array geometries, we show that ADEUPS consistently outperforms task- and array-specific baselines. Finally, we conduct subjective listening tests to confirm the perceptual effectiveness of our approach.
Introduction
Spatial audio conveys the directional and reverberant structure of an acoustic scene and is therefore central to immersive communication, media, and extended-reality applications [2]. Among spatial audio representations, Ambisonics is particularly attractive as it represents the sound field through spherical harmonics independently of the device used to capture it [3]. In practice, however, obtaining a desired Ambisonics representation from microphone recordings is constrained by the physical acquisition system. A microphone array observes the sound field through its geometry and acoustic transfer functions, such that limited spatial sampling, microphone placement, diffraction, and steering-matrix ill-conditioning introduce frequency- and array-dependent distortions [4]. Moreover, the finite spatial resolution of the array geometry allows only a subset of the desired spherical-harmonic orders to be reliably extracted. Consequently, practical Ambisonics acquisition entails two coupled reconstruction tasks: (i) encoding, which recovers the spatial components observed by the microphone measurements while mitigating acquisition-induced distortions; and (ii) upscaling, which infers higher-order components that may not be accurately observable by the array. Achieving high-fidelity Ambisonics from practical microphone arrays thus requires addressing both tasks jointly.
Existing approaches have largely addressed these two components of Ambisonics acquisition separately. For encoding, classical signal-independent methods recover the Ambisonics coefficients by inverting a theoretical or measured array steering model, typically regularized to maintain solution stability [5, 6, 7, 8, 9]. While applicable to different array configurations given their physical models, their accuracy deteriorates when the corresponding inverse problem is ill-conditioned or spatially underdetermined. Signal-dependent parametric approaches instead estimate a compact spatial description of the acoustic scene and use it to resynthesize the desired Ambisonics representation [10, 11]. Such methods provide strong physical interpretability, but rely on accurate estimation of spatial parameters and can degrade in reverberant and multi-source environments. Data-driven encoders seek to overcome these limitations by learning the encoding mapping from examples, either by estimating an array-specific linear transform [12, 13] or by directly mapping microphone recordings to Ambisonics coefficients [14]. Recent approaches improve generalization by conditioning the learned mapping on microphone locations [15] or on measured array transfer functions (ATFs) [16]. Nevertheless, the learned mapping remains coupled to the acquisition interface, e.g., through the assumed microphone count or array-dependent training data, limiting its applicability across arbitrary recording configurations.
A complementary line of work addresses Ambisonics upscaling, where higher-order spatial components are inferred from an available lower-order representation. Classical approaches exploit structural models of the sound field, most commonly through sparse plane-wave decomposition and compressed sensing [17, 18, 19, 20, 21], or through parametric spatial analysis and resynthesis [11, 22]. These techniques can effectively recover higher-order information when the assumed spatial structure holds, but become less reliable in complex reverberant scenes where the sound field is not well represented by a small number of directional components. More recently, data-driven approaches have learned the low-to-high-order mapping using multi-stage and hierarchical neural architectures [23, 24, 25], waveform-domain networks [26, 27], and physics-informed learning [28]. Generative models have also recently been explored for recovering the inherently underdetermined higher-order components [29, 30, 31, 32]. However, these approaches generally operate on already-encoded Ambisonics signals and assume that their observed low-order components provide a reliable representation of the underlying sound field. Consequently, they do not account for the array-dependent distortions, spatial aliasing, and estimation errors introduced during the preceding encoding stage. While encoding and upscaling can be applied separately, array-dependent distortions and reconstruction errors introduced during encoding can be carried into the upscaling stage, whose input may differ from the ideal Ambisonics signals used for its design or training. Rather than learning separate mappings for encoding and upscaling, one can formulate the recovery of an ideal high-order Ambisonics (HOA) representation as a single inverse problem directly from microphone measurements. This motivates a joint approach to encoding and upscaling, recovering the sound field despite partial and distorted observations. Our prior work, ADEPS [1], adopted this approach for Ambisonic signals via Diffusion Posterior Sampling (DPS) [33], focusing primarily on refining spatially observable orders, i.e. encoding. Concurrently, Moliner et al. [34] applied a similar formulation to Ambisonics room impulse responses (ARIRs). However, despite the competitive encoding performance of ADEPS, the recovery of higher-order components degrades drastically as the order increases—both objectively and subjectively. Motivated by this limitation, we propose Ambisonics Diffusion Encoding and Upscaling via Posterior Sampling (ADEUPS), a unified, generative, array-agnostic framework for reconstructing HOA representations directly from microphone-array measurements. To this end, we train a diffusion model on ideal HOA using a novel order-masking strategy, which enforces interchannel dependencies and learns a robust sound-field prior independent of the acquisition hardware. At inference, the microphone configuration is captured by an acoustic forward model that samples the target sound field at the physical locations and re-encodes the resulting signals into Ambisonics. Embedded within DPS, this operator, combined with score blending, steers the diffusion reverse process to yield the reconstructed HOA representation. Together, these two contributions allow the same learned prior to compensate for acquisition-induced distortions while inferring spatial components beyond the array’s resolvable order. This separation between the learned prior and the physical observation model enables zero-shot operation across microphone geometries and channel counts without retraining. Extensive numerical evaluations and subjective listening tests demonstrate its effectiveness across diverse array configurations and reconstruction settings, where it achieves state-of-the-art performance—surpassing array-specific and semi-agnostic baselines.
The main contributions are as follows:
Array-Agnostic Framework: By utilizing linearly encoded Ambisonics derived from the captured microphone signals, and performing refinement and upscaling entirely within the Ambisonic domain, our generative model decouples the sound-field prior from array geometry, generalizing zero-shot to arbitrary microphone counts and configurations without retraining.
Unified Encoding and Upscaling: We solve Ambisonics encoding and order upscaling within a single inverse problem, eliminating error accumulation and performance bottlenecks inherent to cascaded architectures.
Physical Guided Posterior Inference: By leveraging the physical measurement operator within test-time posterior sampling, the ideal generative prior offers inductive bias against measurement noise, spatial aliasing, and physical array imperfections by constraining solutions to the manifold of ideal HOA signals.
Extensive Experimental Validation: We comprehensively evaluate ADEUPS across diverse across diverse microphone geometries and counts, reconstruction orders, considering both encoding and upscaling settings and comparing against dedicated task- and array-specific baselines. In addition to objective spatial and signal-fidelity measures, we conduct subjective listening tests to assess the perceptual quality of the reconstructed Ambisonics signals.
The rest of this paper is organized as follows. § 2 introduces the Ambisonic signal model, establishes the general problem formulation, and provides the mathematical foundation for DPS. Section III details our proposed framework, where we cast the Ambisonic inverse problem explicitly into the DPS sampling process, describe our order-based masking strategy for intra-order learning, and present our blended score approach alongside a tailored noise scheduler. Section IV presents a comprehensive numerical study evaluating our approach across various array configurations for both encoding and upscaling tasks against task-specific baselines. § 5 reports the results of our subjective listening tests. Finally, Section VI concludes the paper.
Throughout this paper, we denote multi-dimensional quantities using boldface notation, with their dimensions specified upon introduction. Calligraphic letters denote sets, e.g., , with and being the sets of real and complex numbers, while , , , , and are the indicator function, norm, Frobenius norm, transpose, conjugate transpose, and Moore-Penrose pseudo-inverse, respectively.
Signal Model and Preliminaries
Signal Model
Ambisonics is a widely used spatial audio representation based on the spherical Fourier transform [3]. In this framework, a continuous sound pressure is decomposed into a series of spherical harmonics (SH) functions of order and degree , weighted by frequency-dependent Ambisonics coefficients , where denotes the wavenumber. When sampling this sound field using an arbitrary array of omnidirectional microphones, where specifies the position of the -th microphone in spherical coordinates [35], the acoustic pressure captured by the -th microphone can be modeled as a linear combination of these coefficients [35].
In (1), denotes the order-dependent radial function, represents the real-valued SH function, and is zero-mean additive measurement noise, assumed to be uncorrelated with the Ambisonics coefficients and i.i.d. across all microphones.
Because the radial functions decay rapidly beyond the , the infinite series in (1) is practically truncated at an effective maximum order [35]. This truncation yields a finite set of coefficients, enabling a compact matrix representation of the array acquisition process:
where stacks the truncated Ambisonics coefficients, is the array noise vector, and is the modal steering matrix. This steering matrix factors into
where is a diagonal matrix containing the radial terms and evaluates the spherical harmonics at the microphone coordinates.
Problem Formulation
Ambisonics reconstruction [7] aims to estimate the SH coefficients from discrete microphone array measurements (2). Accordingly, our objective is to reconstruct a desired order- Ambisonics representation directly from the microphone-array measurements in (2). We represent such a reconstruction by an operator satisfying
where denotes the reconstructed Ambisonics coefficients. Unlike conventional encoding, the target order need not be restricted to the spatial order that can be reliably recovered through direct inversion of the array model. Consequently, (4) encompasses both encoding, namely the recovery of spatial components supported by the microphone observations, and upscaling, namely the inference of higher-order components that are insufficiently observed by the array.
By (3), the mapping (4) can be formulated as an inverse problem [5, 12]. However, recovering from (4) gives rise to several challenges:
C1 Noise Amplification: At low frequencies (), the decay in at higher orders renders direct matrix inversion severely ill-posed, leading to extreme microphone noise amplification.
C2 Non-Uniform Sensitivity: Irregular or non-ideal spatial arrangements of microphones degrade the condition number of , causing direction- and channel-dependent reconstruction errors.
C3 High-Frequency Spatial Aliasing: At frequencies for which the microphone array does not sufficiently sample the sound field, spatial aliasing introduces frequency- and geometry-dependent distortions into the recoverable Ambisonics coefficients.
C4 Underdetermined High-Order Recovery: When the desired Ambisonics order exceeds the spatial dimensionality supported by the microphone observations, e.g., direct reconstruction becomes underdetermined. Conventional encoding must therefore restrict the recovered order, leaving higher-order spatial information unresolved.
C5 Array-Invariance: The encoder should be applicable across array geometries and microphone counts . Although – represent physical degradations in Ambisonic encoding while pertains to Ambisonics upscaling, both stem from the same underlying inverse problem. Our objective is thus to establish a unified spatial mapping operator that addresses these challenges across diverse array configurations (). To this end, we assume access during training to ideal, array-agnostic -order Ambisonic coefficients, formalized as the dataset
Preliminaries
Here we review some basic preliminaries needed to formulate our proposed ADEUPS in the following section. These include conventional linear encoding, as well as some basics in generative diffusion modelling and sampling.
Linear Encoder
1) Linear Encoder: Conventional linear encoding tackles the encoding component of (4) by recovering Ambisonics coefficients directly from the microphone measurements through a frequency-dependent linear operator. For a prescribed order , it forms
Let contain the columns of in (2) associated with Ambisonics coefficients up to order . A common encoding rule is its Tikhonov-regularized pseudoinverse [6, 7]:
where controls the trade-off between inversion accuracy and noise amplification. While this construction applies to any known steering matrix and thus accommodates arbitrary geometries, its output remains constrained by the rank and conditioning of . Consequently, the maximum achievable order is bounded by , which, under , cannot upscale to recover the HOA if . Increasing suppresses unstable modes at the expense of model mismatch and spatial bias, while decreasing it amplifies microphone noise. Linear encoding consequently remains also sensitive to challenges –.
Diffusion Models
2) Diffusion Models: Diffusion models map a complex data distribution to a tractable prior distribution (typically Gaussian) along a continuous trajectory. In the forward process, a clean sample is progressively corrupted across time according to
where defines the noise schedule. To generate samples, this process is reversed by solving the reverse-time stochastic differential equation (SDE) [36]
where is referred to as the drift term, is the diffusion coefficient, is a time-reversed Wiener process, and represents the marginal density of the corrupted state at time . The gradient defines the marginal score function.
In practice, a neural network (NN) is often trained to estimate this score function. Under the Variance Exploding (VE) formulation of elucidating diffusion models (EDMs) [37], the drift term is removed () and the diffusion parameter is set to , which directly yields . EDMs precondition the NN as a denoiser according to
where is the model to be trained, and the coefficients are given by
Here, represents the standard deviation of the data distribution, while provides a log-scaled noise embedding. These coefficients normalize the network inputs and output targets across noise levels while preserving a skip connection to the data manifold when .
Via Tweedie’s formula [38], the marginal score function is related to the ideal denoiser by
Sample generation can thus be executed by solving the probability flow ordinary differential equation (ODE) [39]
or stochastically by integrating an equivalent SDE solver.
Diffusion Posterior Sampling
Rather than generating samples, this approach can be used to solve inverse problems. Consider an unknown clean sample is observed via
where denotes a known, potentially nonlinear degradation operator. Sampling from the posterior distribution can be achieved by substituting the posterior score function into the probability flow ODE in (16). By applying Bayes’ rule, the posterior score decomposes into
where the first term is supplied by the unconditional diffusion prior and the second term corresponds to the likelihood score which we denote as .
Because the exact likelihood is intractable for intermediate states , it is commonly approximated via the estimated target [33], [40]:
where follows from the Gaussian observation model in (17) and is a time-dependent weight controlling the guidance strength. Substituting this approximation together with (15) into the probability flow ODE yields the guided generative process:
This gradient steering pushes the sampling trajectory toward reconstructions that remain both realistic under the prior and faithful to the observation .
ADEUPS
We next introduce ADEUPS, which formulates the inverse problem in the Ambisonics domain to enable zero-shot encoding and upscaling for arbitrary array configurations using a generative prior. Subsection III-A frames this inverse problem within the DPS framework, Subsection III-C outlines our training scheme with order masking and conditioning, and Subsection III-B details the inference pipeline with scheduled score blending, followed by a discussion in Subsection III-D. All signals are defined in the short-time Fourier transform (STFT) domain; time-frame and frequency-bin indices are omitted for brevity.
DPS Formulation for Ambisonics Reconstruction
Recall that due to C4, the desired Ambisonics order, which is the order of the Ambisonics prior, holds . Similar to [41], [42], we define our prior in the magnitude-compressed STFT domain. We denote this element-wise compression operator as , where are hyperparameters and denotes the phase angle.
We formulate our inverse problem directly in the Ambisonics domain by defining the observation vector as the compressed encoded signal of order (matching the prior order). Specifically, writing the observations of (17) as and substituting the signal acquisition model (2) yields:
Accordingly, we define the forward operator in (17), which maps the ideal compressed order sample to the corresponding observation domain, as
This operator explicitly captures the sequence of linear STFT decompression, spatial sampling by microphones, Ambisonics linear encoding, and STFT compression. By modeling these array-specific degradation artifacts (C1–C4) exclusively within the forward pipeline at inference time, the resulting DPS formulation, once properly trained, supports zero-shot generalization across arbitrary microphone geometries (C5).
Inference
We next describe the inference procedure assuming that the parameters of the order-conditioned denoiser are given; their training is discussed in Subsection III-C. The conditioning variable specifies the highest Ambisonics order provided to the denoiser as trusted context. For an array with microphones, we set this order to .
Recall that DPS reconstructs the unknown signal by replacing the unconditional score in the reverse diffusion dynamics with the posterior score in (18). Accordingly, ADEUPS inference follows (20) using
where the prior score promotes consistency with the learned distribution of ideal Ambisonics signals, while the likelihood score enforces consistency with the array observation through the physical forward operator (22). We next specify these two terms.
Order-conditioned prior score: A single full-order diffusion score does not explicitly distinguish between Ambisonics components that can be supported by the array and higher-order components that must be inferred from their lower-order context. We therefore exploit two complementary operating modes of the same order-conditioned denoiser. Evaluating it with yields the generation score
which uses the lower-order Ambisonics components as trusted context for generating the remaining HOA components. Evaluating the denoiser with gives the refinement score
which operates on the complete order- representation and promotes full-order denoising and refinement. The mechanism by which these two behaviors are induced during training is detailed in Subsection III-C.
The relative importance of generation and refinement varies along the reverse diffusion trajectory. At high noise levels, the higher-order structure has not yet been established, making generation from the trusted lower-order context particularly important. At low noise levels, the reconstructed HOA structure is already largely formed, and continued generative guidance can disturb its fine details. We therefore employ score blending [43] and define
where control the relative score contributions, and is a threshold hyperparameter which is set empirically. Thus, the early stages of sampling jointly establish higher-order structure and refine the current Ambisonics estimate, whereas the final stages (where the noise level is below the threshold) focus on full-order refinement.
Array-dependent likelihood score: Having formed the prior score, we instantiate the likelihood term of the generic DPS construction in (19). Following Tweedie’s relation in (15), the blended score provides the clean-signal proxy
The corresponding likelihood score is
where is defined in (22). Following [40], [44], we normalize the guidance strength according to
where is a constant scaling parameter. Consequently, the specific microphone configuration affects the posterior trajectory only through in the likelihood score, while the learned prior score remains independent of the acquisition array.
Posterior sampling: Substituting (26) and (28) into the DPS dynamics in (20) yields the ADEUPS reverse trajectory. We discretize it over steps using the -spaced noise schedule of [37],
for , with . We initialize the trajectory around the encoded observation as
where . We then use the first-order Euler sampler with stochastic churn [37]. Denoting the churn-perturbed state and noise level at iteration by and , respectively, the posterior scores are evaluated at and the state is updated as
After the final sampling step, applying yields the reconstructed order- Ambisonics coefficients. The complete inference procedure, including the stochastic churn operations, is summarized in Algorithm 1 and illustrated in Figure 1.

Figure 1. ADEUPS Pipeline
Algorithm 1: ADEUPS Inference
Input: Signals , Steering matrix , Denoiser , Hyperparameters , Churn parameters
Set and via (7)
Compute
Sample
for to do /* Stochastic Churn sampling */
Sample
/* Calculate refinement score */
/* Blend score if over threshold */
if then
else
/* Perform DPS step */
Update
return
Order-Masking Prior Training
The inference procedure in Subsection III-B assumes access to an order-conditioned denoiser that can operate both as a full-order refinement prior and as a conditional generator from a prescribed trusted order . We next describe how these complementary behaviors are induced during training.
Training is carried out exclusively using ideal order- Ambisonics signals from the dataset [5]; no microphone signals, steering matrices, or array-dependent encoding operators are used. The resulting learned prior is therefore independent of the acquisition configuration.
To expose the denoiser to different amounts of available spatial information, we randomly mask complete Ambisonics orders during training. Let denote the Ambisonics channel index and let denote its corresponding SH order. For each training sample, we draw a trusted order according to a categorical distribution with probabilities , and define
with the indicator function and with the corresponding channel-wise mask whose entries are . Thus, all components up to order are treated as trusted, while those of orders are regarded as unavailable.
Given an ideal sample , a noise level , and , we construct the input to the denoiser channel-wise as
Consequently, trusted channels follow the standard diffusion corruption process in (8), whereas channels above order contain only noise. The training target, however, always remains the complete clean order- Ambisonics signal . Hence, when , successful denoising requires the network to infer the masked higher-order components from the available lower-order context, thereby learning the inter-order dependencies required by the generation score in (24). For , no channels are masked and the model reduces to full-order denoising, as required by the refinement score in (25).
The standard EDM preconditioning in (10) must also be adapted to this masking operation. In particular, directly retaining the skip connection on a masked channel would propagate its pure-noise input to the denoiser output. We therefore define the order-conditioned denoiser as
where and are given in (11), while
For a trusted channel, i.e., , these expressions reduce to the standard EDM coefficients in (11). For a masked channel, , the skip coefficient vanishes, preventing its pure-noise realization from being directly propagated to the output, while provides the appropriate scaling for predicting the missing component through . The trusted order is additionally provided explicitly to , allowing the same network to distinguish between the different order-availability patterns.
Accordingly, the parameters are learned by minimizing the channel-wise preconditioned denoising objective
where is the number of Ambisonics channels, and denote the numbers of frequency bins and time frames, respectively, denotes a sampled diffusion noise level, and vector division is taken element-wise. Through the joint sampling of and , a single array-independent denoiser thus learns both behaviors required at inference: for it models the full order- Ambisonics distribution, while for it learns to complete the higher-order representation from the trusted lower-order components.
Discussion
The proposed ADEUPS construction is tailored to the coupled challenges in Subsection III-B by separating the information supplied by the learned Ambisonics prior from that supplied by the physical acquisition model. In particular, rather than relying solely on inversion of the possibly ill-conditioned steering matrix, the posterior trajectory combines a learned prior over ideal Ambisonics with measurement consistency through . This provides a data-driven regularization mechanism for the noise amplification and non-uniform sensitivity in –, while explicitly accounting for the array-dependent acquisition effects underlying spatial aliasing in C3. The order-conditioned generation score exploits learned inter-order dependencies to infer components that are insufficiently determined by the measurements, thereby addressing the high-order recovery challenge in C4, with the likelihood score maintaining consistency with the available physical observations. Crucially, the diffusion prior is trained only on ideal Ambisonics signals, whereas the steering and encoding matrices enter solely through the test-time forward operator. Consequently, a single learned model can be applied to different microphone geometries and channel counts without retraining, addressing C5.
The inference procedure introduces several parameters governing the relative strength and numerical realization of posterior sampling. The coefficients , , and balance generation and refinement, determines when the sampler transitions to refinement-only operation, and controls the strength of measurement guidance. The noise schedule, number of sampling steps, and stochastic-Churn parameters further determine the numerical accuracy and computational cost of the reverse trajectory, while the distribution used during training determines the trusted-order configurations to which the denoiser is exposed. These quantities can be selected on held-out validation data and subsequently kept fixed across array configurations. In contrast, the target order is specified by the reconstruction problem rather than being an algorithmic hyperparameter. Its relation to the effective acquisition order nevertheless affects the fidelity of the forward model: when , contributions from orders above are omitted from (22), resulting in model mismatch due to unmodeled spatial components. The impact of this mismatch is examined empirically in Subsection IV-C3.
The formulation also suggests several directions for further development. First, the current trusted order is determined from the microphone count, whereas in practice observability also depends on frequency, array geometry, and the conditioning of the corresponding steering matrix. Replacing this hard order threshold by frequency- and array-dependent confidence information, potentially through soft order or channel masks, could more accurately distinguish measured from prior-dominated components. Further, the present likelihood assumes that the array steering matrix is accurately known; extending posterior sampling to account for calibration errors or uncertainty in the physical acquisition model would broaden its applicability to practical recording systems. These extensions are left for future study.
Empirical Evaluation
Here, we numerically evaluate ADEUPS1, and compare it with several Ambisonics reconstruction methods. The evaluation is structured as follows: Subsection IV-A outlines the experimental setup and the datasets used for training and evaluation. Subsection IV-B details the methodology, including the model configurations, baselines, and the evaluation metrics, while Subsection IV-C reports the numerical results.
Experimental Setup
Training Data
Training Data: For training, we use order ARIRs sampled at 16 kHz, generated using HARP room simulation code [45]. We randomly generate rooms with dimensions m and reverberation times s. The Ambisonics measurement point is randomly positioned within a m region on the -plane at a height of 1.2 to 1.5 m along the -axis. Point sources are positioned with distances to the Ambisonics measurement point randomly sampled from 0.8 to 1.5 m, with direction-of-arrivals (DOAs) also drawn randomly. In total, we generate 2000 rooms with 10 ARIRs per room, yielding 20,000 ARIRs. For the speech dataset, we use VCTK [46], which consists of approximately 44 hours of data. We employ an on-the-fly data loader that randomly crops 2.04 s speech segments per source at runtime and convolves them with the corresponding ARIRs. The number of active speakers is randomly chosen to be 1 or 2.
Evaluation Data
Evaluation Data: To evaluate the models, we generate 100 acoustic scenes following the training framework, with two key differences: (i) the Ambisonics order is set to ; and (ii) speech signals are taken from the WSJ0 dataset [47]. We evaluate performance across 13 test array geometries: the 7-microphone Project Aria glasses [48] alongside 4 irregular realizations for each microphone count . For Project Aria, we utilize its measured steering matrix, whereas for the irregular geometries, steering matrices are computed analytically. Following [15], the microphone positions for the irregular arrays are randomly sampled subject to minimum and maximum pairwise distances of 0.02 m and 0.18 m, respectively. Microphone signals are generated by projecting the ground-truth Ambisonics onto each array geometry via (2) and adding white Gaussian noise at an signal-to-noise ratio (SNR) of 50 dB, totaling in 1300 microphone signals.
Methodology
Benchmarks
Methodology
We base our diffusion model on the NCSN++M [49] backbone. We incorporate a fully connected layer for trusted order conditioning, replace additive conditioning with adaptive linear norm [50], and resize the input and output fully connected layers to channels, resulting in a model with 31.1 M parameters.
For the STFT, an FFT length of 510 samples, a hop size of 128 samples, and a Hann window were employed, resulting in 256 frequency bins and 256 time frames. The compression parameters were set to and to achieve a unit standard deviation (). During training, the noise schedule is configured with , , and . The order-masking probability is set to . During inference, we adopt the same parameters except for , which is reduced to 4. Sampling is performed with steps, resulting in 150 function evaluations, using stochastic churn parameters , , , and . The score weights are set to , , and , with a guidance threshold of .
Benchmarks: We compare our framework against both model-based methods and learning-based methods, with the former being array-agnostic by design. To the best of our knowledge, no existing data-driven method supports varying array configurations and sizes, and we thus benchmark against both semi-agnostic and array-specific approaches:
Linear Encoder: The encoder defined in (7).
Parametric Method: Based on [10], we estimate the direct and reverberant components of the sound field and
resynthesize the Ambisonics signal up to the target order. To establish an upper bound, we use oracle DOAs to estimate direction of sound sources.
UNet-Encoder: Based on [12], this approach utilizes a U-Net architecture to estimate the encoding matrix . As it is array-specific, we train it on one of the four-microphone arrays from the test set. The network has a size of 7.9M parameters.
Gen-A: Based on [15], we employ a U-Net NN that estimates the encoding matrix conditioned on the microphone positions to encode up to the spatially resolvable order. Because the NN is constrained to a fixed channel count, we train it across diverse array geometries with randomly positioned microphones. This network is of size 10.2M parameters.
Diff-Encoder: We train a diffusion model using the NCSN++M architecture to generate an -order Ambisonics signal conditioned on the captured microphone signals . Conditioning is performed by concatenating with the corrupted state along the channel dimension. This approach is array-specific, and is trained on one of the four-microphone arrays from the test set. The network size in this case is 30.8M parameters.
Diff-Upscaler: Inspired by [32], we utilize the NCSN++M architecture to upscale from first-order Ambisonics (FOA) to order . We train the model on FOA signals linearly encoded from randomly positioned arrays with microphones. This network is of size 30.8M parameters.
A summary of the models used in the evaluation is presented in Table I.
Evaluation Metrics
2) Evaluation Metrics: Let and denote the reference and reconstructed time-domain Ambisonic signals, respectively. Furthermore, let and represent their uncompressed STFT coefficients for channel , with and denoting the time-frame and frequency-bin indices, respectively. We evaluate performance using the following metrics:
(i) Log-Magnitude Spectral Error: Measures the spectral magnitude fidelity across all channels , time frames , and frequency bins as
(ii) Mean Spectral Coherence: Evaluates the average normalized cross-spectral coherence between the reference and estimated signals:
(iii) Signal Invariant Signal Distortion Ratio (SI-SDR): Evaluates overall time-domain fidelity while remaining invariant to global gain mismatches via a scalar projection
Results
We present the results in three parts. First, we compare the proposed method to existing array-agnostic baselines, followed by a comparison against array-specific methods. Finally, we conduct an ablation study to analyze our key design choices.
Full- and Semi-Agnostic Methods Evaluation
1) Full- and Semi-Agnostic Methods Evaluation: Table II presents performance of the array-agnostic methods across different orders and array configurations. For first-order encoding, we compare both model-based approaches with ADEUPS and Gen-A (which is restricted to five-microphone arrays). Our method outperforms all baselines in SI-SDR by a substantial margin and achieves superior spectral error across nearly all configurations, while remaining second-best in coherence.
For orders 2-5, the linear encoder and Gen-A are inapplicable; thus, we also evaluate Diff-Upscaler, which operates on linearly encoded FOA. Because its input lies in the Ambisonics domain rather than the microphone domain, it is array-agnostic in the sense that varying microphone counts do not necessitate retraining. We trained Diff-Upscaler exclusively on four- and six-microphone arrays to evaluate its behavior on two out-of-distribution scenarios. On FOA derived from a five-microphone array, the upscaler performs comparably to the four- and six-microphone cases; this demonstrates that although it was not trained on these specific spatial aliasing artifacts, they closely resemble those of the training arrays, which share an open-sphere topology. Conversely, on the Aria glasses array, the upscaler fails entirely. This highlights a fundamental pitfall of supervised training on specific array configurations: generalization becomes unreliable for geometries that deviate significantly from the training set.
These results highlight the strength of ADEUPS, which achieves superior SI-SDR across all array configurations without having observed any of them during training. This performance gap between diffusion models of identical capacity stems from the underlying distribution being learned: ADEUPS trains on a clean -order prior and incorporates array-specific forward operators solely during inference, thereby simplifying optimization. In contrast, the diffusion upscaler must learn a joint distribution entangled with diverse encoding artifacts, substantially increasing training complexity and hindering generalizability.
Array Specific Methods Evaluation
2) Array Specific Methods Evaluation: We next evaluate ADEUPS against array-specific baselines: Diff-Encoder, which directly maps microphone signals to order Ambisonics signals, and UNet-Encoder, which estimates an encoding matrix to project the microphone signals to order Ambisonics. The evaluation results on a four-microphone2 open array are summarized in Table III.
For first-order Ambisonics, our method outperforms both baselines in SI-SDR, whereas the UNet-Encoder achieves the best performance in terms of coherence and spectral error. For orders 2 through 5, we compare directly with Diff-Encoder; our method achieves higher SI-SDR and comparable coherence, while underperforming the baseline by only a small margin in spectral error. Overall, these results demonstrate the robustness and flexibility of our approach compared to methods trained on a specific geometry. Crucially, when evaluated on an unseen 4-microphone configuration, the baselines fail entirely, whereas our method maintains consistent performance.
| Method | Parameters | Input Signal | Output Order | Supported Arrays |
| Linear Encoder | 0 | Microphones | Arbitrary | |
| Parametric | 0 | Microphones | Arbitrary | |
| UNet-Encoder | 7.9M | Microphones | Fixed geometry () | |
| Gen-A | 10.2M | Microphones | Arbitrary geometry () | |
| Diff-Encoder | 30.8M | Microphones | Fixed geometry () | |
| Diff-Upscaler | 30.8M | FOA | Arbitrary | |
| ADEUPS (ours) | 31.1M | Microphones | Arbitrary |
Table 1. Summary of Models and their Specifications.
| Order | Method | SI-SDR [dB] ↑ 4 Mics | SI-SDR [dB] ↑ 5 Mics | SI-SDR [dB] ↑ 6 Mics | SI-SDR [dB] ↑ Aria | Mean Spec. Error [dB] ↓ 4 Mics | Mean Spec. Error [dB] ↓ 5 Mics | Mean Spec. Error [dB] ↓ 6 Mics | Mean Spec. Error [dB] ↓ Aria | Coherence ↑ 4 Mics | Coherence ↑ 5 Mics | Coherence ↑ 6 Mics | Coherence ↑ Aria |
| 1 | Linear | 6.54 | 7.63 | 8.86 | 3.40 | 9.50 | 9.28 | 8.33 | 10.9 | 0.82 | 0.82 | 0.83 | 0.76 |
| 1 | Param. (oracle) | 3.73 | 3.67 | 3.77 | 2.28 | 5.83 | 5.87 | 5.85 | 6.22 | 0.71 | 0.70 | 0.71 | 0.67 |
| 1 | Gen-A | - | 3.75 | - | - | - | 7.37 | - | - | - | 0.78 | - | - |
| 1 | ADEUPS | 10.52 | 11.27 | 13.13 | 6.44 | 5.61 | 5.65 | 5.15 | 7.42 | 0.79 | 0.79 | 0.83 | 0.66 |
| 2 | Param. (oracle) | 2.41 | 2.55 | 2.60 | 1.50 | 6.27 | 6.25 | 6.23 | 6.53 | 0.69 | 0.68 | 0.68 | 0.64 |
| 2 | Diff-Upscaler | 5.71 | 6.19 | 6.74 | -3.36 | 9.20 | 9.03 | 8.40 | 20.18 | 0.74 | 0.74 | 0.76 | 0.63 |
| 2 | ADEUPS | 8.34 | 9.49 | 11.04 | 5.87 | 6.33 | 6.28 | 5.84 | 7.86 | 0.76 | 0.77 | 0.81 | 0.65 |
| 3 | Param. (oracle) | 1.80 | 2.00 | 2.00 | 1.00 | 6.47 | 6.44 | 6.41 | 6.69 | 0.67 | 0.66 | 0.67 | 0.63 |
| 3 | Diff-Upscaler | 5.30 | 5.70 | 6.13 | -6.60 | 9.18 | 9.02 | 8.50 | 20.95 | 0.70 | 0.70 | 0.72 | 0.63 |
| 3 | ADEUPS | 7.58 | 8.59 | 9.69 | 5.29 | 6.67 | 6.62 | 6.23 | 8.09 | 0.74 | 0.75 | 0.79 | 0.60 |
| 4 | Param. (oracle) | 1.42 | 1.62 | 1.64 | 0.70 | 6.65 | 6.61 | 6.58 | 6.82 | 0.66 | 0.65 | 0.66 | 0.62 |
| 4 | Diff-Upscaler | 4.79 | 5.26 | 5.64 | -9.13 | 9.21 | 9.05 | 8.60 | 20.26 | 0.68 | 0.68 | 0.69 | 0.54 |
| 4 | ADEUPS | 6.99 | 7.94 | 8.82 | 4.70 | 6.99 | 6.95 | 6.58 | 8.33 | 0.73 | 0.74 | 0.77 | 0.62 |
| 5 | Param. (oracle) | 1.78 | 1.39 | 1.40 | 0.50 | 6.77 | 6.71 | 6.67 | 6.91 | 0.65 | 0.64 | 0.65 | 0.61 |
| 5 | Diff-Upscaler | 4.22 | 4.82 | 5.18 | -11.23 | 9.26 | 9.09 | 8.67 | 18.82 | 0.65 | 0.65 | 0.67 | 0.52 |
| 5 | ADEUPS | 6.34 | 7.23 | 8.00 | 4.02 | 7.24 | 7.20 | 7.42 | 8.52 | 0.71 | 0.72 | 0.76 | 0.61 |
Table 2. Performance Comparison Across Orders and Arrays (bold indicates best performance).
| Order | Method | SI-SDR [dB] ↑ | MSE [dB] ↓ | Coherence ↑ |
| 1 | UNet-Encoder | 6.65 | 5.73 | 0.85 |
| 1 | Diff-Encoder | 9.60 | 5.98 | 0.78 |
| 1 | ADEUPS | 10.83 | 5.81 | 0.77 |
| 2 | Diff-Encoder | 7.60 | 6.34 | 0.74 |
| 2 | ADEUPS | 8.42 | 6.45 | 0.76 |
| 3 | Diff-Encoder | 7.03 | 6.57 | 0.72 |
| 3 | ADEUPS | 7.72 | 6.75 | 0.74 |
| 4 | Diff-Encoder | 6.52 | 6.78 | 0.70 |
| 4 | ADEUPS | 7.17 | 7.05 | 0.72 |
| 5 | Diff-Encoder | 6.01 | 7.00 | 0.68 |
| 5 | ADEUPS | 6.52 | 7.27 | 0.71 |
Table 3. Performance Comparison for a Specific 4 Mic Array2.
Ablation Study
To quantify the individual contributions of our proposed design choices, we conduct an ablation study on the evaluation dataset, reporting the mean metrics across all microphone arrays for each Ambisonics order. Table IV summarizes the ablation sequence where the sampling steps where chosen such that each method had 150 model evaluations.
| Method/Order | SI-SDR [dB] ↑ 1 | 2 | 3 | 4 | 5 | Mean Spec. Error [dB] ↓ 1 | 2 | 3 | 4 | 5 | Coherence ↑ 1 | 2 | 3 | 4 | 5 |
| ADEPS | 10.53 | 5.79 | 3.87 | 2.52 | 1.30 | 5.48 | 6.57 | 7.47 | 8.31 | 9.16 | 0.76 | 0.72 | 0.69 | 0.66 | 0.64 |
| ADEPS + Masking | 10.38 | 7.47 | 5.91 | 4.64 | 3.34 | 5.52 | 6.24 | 6.77 | 7.30 | 7.93 | 0.77 | 0.74 | 0.71 | 0.69 | 0.67 |
| ADEPS + Masking + Ref Cond. | 10.58 | 7.29 | 5.65 | 4.39 | 3.20 | 5.42 | 6.19 | 6.79 | 7.36 | 8.00 | 0.79 | 0.75 | 0.73 | 0.70 | 0.68 |
| ADEPS + Masking + Gen Cond. | 0.40 | 2.70 | 3.88 | 4.40 | 4.47 | 12.89 | 10.09 | 9.12 | 8.90 | 8.83 | 0.59 | 0.68 | 0.70 | 0.71 | 0.70 |
| ADEPS + Masking + Score Blending | 10.04 | 4.40 | 3.04 | 2.62 | 1.70 | 6.68 | 10.24 | 11.60 | 12.58 | 13.05 | 0.75 | 0.73 | 0.71 | 0.70 | 0.68 |
| ADEPS + Masking + Score Blending + Scheduler (ADEUPS) | 11.24 | 9.33 | 8.37 | 7.67 | 6.95 | 5.61 | 6.28 | 6.62 | 6.95 | 7.20 | 0.79 | 0.77 | 0.75 | 0.74 | 0.72 |
Table IV. Ablation study evaluating the incremental contributions of order-masking and score-blending. The baseline configuration corresponds to ADEPS [1] (without masking or conditioning, using a first-order stochastic Euler sampler). All configurations are evaluated under an identical budget of 150 model evaluations. Best results are shown in bold.
We define the baseline as our previous framework, ADEPS [1], which performs Ambisonics encoding via DPS using an EDM-based learned prior with a first-order stochastic Euler sampler, omitting both order-masking during training and score-blending during inference. Incorporating HOA masking into training—even without conditioning—substantially improves the SI-SDR for higher-order components while maintaining comparable performance on lower-order ones.
Next, we evaluate models trained with both masking and conditioning, examining the individual effects of conditioning on either the refinement score or the generation score. Conditioning solely on the refinement score yields results similar to the masking-only framework. In contrast, using only the generation score severely degrades performance at lower orders due to the absence of a skip connection. Notably, unlike the other configurations, the generation-only model exhibits a reversed performance trend, where performance improves with higher Ambisonics orders.
This observation motivates a hybrid approach that leverages the strong low-order performance of the refinement score while utilizing the low-order components to guide higher-order generation. However, naive score blending initially yields suboptimal results because the generation score degrades quality at lower noise levels. At these noise levels, high-order components are already established; attempting to generate them primarily from low-order components becomes ineffective and even detrimental. To address this, we empirically define a threshold , below which the sampler transitions from score blending exclusively to the refinement score. Figure 2 illustrates the iteration-wise dynamics with and without the scheduler—evaluated on the average SI-SDR for both low-order and high-order components—demonstrating its effectiveness in stabilizing convergence during sampling.

Figure 2. SI-SDR over time for ADEUPS with and without scheduler.
We next compare performance across different values of , evaluating our test set at . Table V reports the results per Ambisonics order, averaged across all microphone arrays. At , we observe peak performance because , rendering our forward measurement model (19) exact and thus establishing the upper bound on performance. Beyond this, we observe a noticeable performance drop from to , but only a minimal change between and . These findings confirm two key insights: (i) The initial performance degradation stems from the truncated prior order (). Because the prior cannot represent components beyond , it fails to account for energy folding caused by spatial aliasing from higher orders. This behavior is further corroborated by the spectral analysis in Figure 3, which reveals that the performance discrepancy between and is concentrated at high frequencies where spatial aliasing is most pronounced; improving performance under such conditions would thus require a higher-order prior. (ii) The marginal difference between and demonstrates that energy decays rapidly as the order increases. This indicates that serves as a valid practical approximation of the infinite sum for speech data, supporting the core assumption of our framework.

Figure 3. Spectral error and coherence for truncation orders and .
Listening Experiment
While Section IV provides objective evaluations, subjective listening experiments remain essential to capture perceptual attributes beyond the scope of numerical metrics. The experimental setup is detailed in § 5-A, followed by the evaluation methodology in § 5-B. Finally, the subjective results are discussed in § 5-C.
Setup
We follow the framework outlined in Section IV-A2 to generate the evaluation data, setting the dry signal duration to 5 seconds. The listening test evaluates three screen configurations across distinct acoustic and array setups:
Single-speaker FOA: Room dimensions 6.8 8.2 3.0$ m (RT = 380 ms, DRR = dB), using a 4-microphone array centered at (2.9, 6.2, 1.7) m. A male speaker is positioned at a distance of 1.3 m with .
Dual-speaker FOA: Same acoustic environment as (RT = 380 ms, DRR = dB), using a 5-microphone array centered at (2.9, 6.2, 1.7) m. Two sources are positioned in the room: a male speaker located at 1.3 m with and a female speaker at 1.2 m with .
Single-speaker HOA4: Room dimensions 8.8 8.2 3.0$ m (RT = 450 ms, DRR = 3.2 dB), using a 6-microphone array centered at (3.8, 6.2, 1.7) m. A female speaker is positioned at a distance of 1.3 m with .
For each scenario, four stimuli are generated: (i) the ground-truth Ambisonic signal at the target order (FOA for –, HOA of order 4 for ); (ii) a linearly encoded FOA signal; (iii) the target-order Ambisonics signal produced by ADEUPS; and (iv) an Ambisonics signal generated by a data-driven baseline. Because existing deep learning baselines cannot simultaneously accommodate arbitrary array geometries and higher target orders, different models are used across scenarios: UNet-Encoder and Gen-A (which take raw microphone signals as input) for and , and Diff-Upscaler (which operates on the linearly encoded FOA) for .
Methodology
The subjective evaluation was conducted following the Multiple Stimuli with Hidden Reference and Anchor (MUSHRA) methodology [51]. Participants were instructed to rate the overall quality of the stimuli relative to an explicit known reference on a continuous scale from 0 to 100, where 100 denotes no perceptible difference from the reference and lower scores indicate increasing degradation. In each scenario, the ground-truth Ambisonic signal served as the reference (and hidden reference), while the linearly encoded FOA signal served as a low-anchor.
| SI-SDR [dB] ↑ | Mean Spec. Error [dB] ↓ | Coherence ↑ | |||||||||||||
| 1 | 2 | 3 | 4 | 5 | 1 | 2 | 3 | 4 | 5 | 1 | 2 | 3 | 4 | 5 | |
| 5 | 14.68 | 11.27 | 9.92 | 9.12 | 8.37 | 5.05 | 5.77 | 6.14 | 6.47 | 6.73 | 0.85 | 0.83 | 0.81 | 0.80 | 0.78 |
| 10 | 11.29 | 9.38 | 8.41 | 7.73 | 7.01 | 5.60 | 6.27 | 6.62 | 6.95 | 7.20 | 0.80 | 0.78 | 0.76 | 0.74 | 0.72 |
| 15 | 11.24 | 9.33 | 8.37 | 7.67 | 6.95 | 5.61 | 6.28 | 6.62 | 6.95 | 7.20 | 0.79 | 0.77 | 0.75 | 0.74 | 0.72 |
Table 5. Effect of on Performance per Order.
A total of 11 listeners participated in the experiment. Prior to testing, each subject completed a brief training session to familiarize themselves with the range of sound quality and the test interface. The presentation order of the test screens and stimuli within each screen was randomized to mitigate potential bias.
Results
The MUSHRA test results are illustrated in Figure 4. For each test screen, a two-way repeated-measures analysis of variance (ANOVA) was conducted to evaluate the impact of the processing method on subjective quality ratings.

Figure 4. MUSHRA results across test screens: Screen 1 (4-microphone array), Screen 2 (5-microphone array), and Screen 3 (6-microphone array). Horizontal lines denote medians, and diamonds denote means.
Screen 1 (): The analysis revealed a statistically significant main effect for the processing method (, , ), indicating that method choice substantially influenced perceived quality. Post-hoc pairwise comparisons with Bonferroni corrections were performed across the ADEUPS, ground-truth FOA, and UNet-Encoder conditions, which achieved mean scores of 100, 88.72, and 70.90, respectively. No statistically significant difference was observed between ADEUPS and either the ground-truth signal (, ) or the UNet-Encoder (, ). However, ratings for the ground-truth signal were significantly higher than those for the UNet-Encoder (, ).
Screen 2 (): Mauchly’s test indicated a violation of sphericity (, ), requiring Greenhouse-Geisser corrections to the degrees of freedom. The main effect of the processing method remained statistically significant (, , ). Mean scores for ADEUPS, ground-truth FOA, and the UNet-Encoder were 100, 96.09, and 62.68, respectively. Post-hoc Bonferroni-corrected comparisons revealed no significant difference between ADEUPS and the ground-truth signal (, ), whereas Gen-A performed significantly worse than both ADEUPS (, ) and the ground truth (, ).
Screen 3 (): Similarly, a highly significant main effect of the processing method was observed (, , ). The mean scores for ADEUPS, ground-truth FOA, and the Diff-Upscaler were 100, 91.45, and 81.45, respectively. Post-hoc comparisons with Bonferroni correction showed no statistically significant difference between ADEUPS and either the ground-truth signal or the Diff-Upscaler; however, the ground-truth signal rated significantly higher than the Diff-Upscaler.
Conclusions
In this work, we introduced ADEUPS, a novel, array-agnostic framework for Ambisonic encoding and spatial upscaling. By formulating the reconstruction task as a diffusion-based inverse problem solved via DPS, our model learns an acquisition-independent sound-field prior of ideal HOA signals. At inference, array-specific configurations are incorporated into the forward operator to invert device-induced physical artifacts. To further enhance reconstruction quality, we introduced two key innovations: order-masking during training to enforce interchannel dependencies, and score-blending during inference to guide sampling trajectory. Objective and subjective evaluations demonstrate that ADEUPS achieves performance comparable to—and in many cases exceeding—both array-agnostic model-based baselines and data-driven methods trained on specific array topologies. Crucially, ADEUPS operates in a zero-shot manner, requiring no exposure to specific array geometries during training. Future work will explore real-time algorithmic optimizations and extend the framework to dynamic acoustic scenes.
References
- [1]A. Milstein, N. Shlezinger, and B. Rafaely, “Array-agnostic ambisonics encoding via diffusion posterior sampling,” arXiv preprint arXiv:2608.24558, 2026.arxiv.org/abs/2608.24558
- [2]H. Hachabiboglu et al., “Perceptual spatial audio recording, simulation, and rendering: An overview of spatial-audio techniques based on psychoacoustics,” vol. 34, no. 3, pp. 36–54, 2017.
- [3]F. Zotter and M. Frank, Ambisonics: A practical 3D audio theory for recording, studio production, sound reinforcement, and virtual reality. Springer Nature, 2019.DOI
- [4]B. Rafaely, “Analysis and design of spherical microphone arrays,” vol. 13, no. 1, pp. 135–143, 2005.DOI
- [5]A. Politis and H. Gamper, “Comparing modeled and measurement-based spherical harmonic encoding filters for spherical microphone arrays,” in IEEE Workshop on Applications of Signal Processing to Audio and Acoustics (WASPAA), 2017, pp. 224–228.DOI
- [6]S. Moreau, J. Daniel, and S. Bertet, “3d sound field recording with higher order ambisonics—objective measurements and validation of a 4th order spherical microphone,” in 120th Convention of the AES, 2006, pp. 20–23.
- [7]M. A. Poletti, “Three-dimensional surround sound systems based on spherical harmonics,” Journal of the Audio Engineering Society, vol. 53, no. 11, pp. 1004–1025, 2005.
- [8]A. Bastine, L. Birnie, T. D. Abhayapala, P. Samarasinghe, and V. Tourbain, “Ambisonics capture using microphone arrays on head-worn device of arbitrary geometry,” in 2022 30th European Signal Processing Conference (EUSIPCO). IEEE, 2022, pp. 309–313.DOI
- [9]Y. Gayer, V. Tourbain, Z. Ben-Hur, D. Alon, and B. Rafaely, “Ambisonics encoder for wearable array with improved binaural reproduction,” arXiv preprint arXiv:2507.04108, 2025.arxiv.org/abs/2507.04108
- [10]L. McCormack, A. Politis, R. Gonzalez, T. Lokki, and V. Pulkki, “Parametric ambisonic encoding of arbitrary microphone arrays,” vol. 30, pp. 2062–2075, 2022.
- [11]A. Politis, S. Tervo, and V. Pulkki, “COMPASS: Coding and multidirectional parameterization of ambisonic sound scenes,” in IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2018, pp. 6082–6086.
- [12]M. Heikkinen, A. Politis, and T. Virtanen, “Neural ambisonics encoding for compact irregular microphone arrays,” in IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2024, pp. 701–705.arxiv.org/abs/2401.05916
- [13]T. Deppisch, Y. Gao, M. Mittal, B. Stahl, C. Hold, D. Alon, and Z. Ben-Hur, “Residual learning for neural ambisonics encoders,” arXiv preprint arXiv:2601.18322, 2026.arxiv.org/abs/2601.18322
- [14]Y. Qiao, V. Kothapally, M. Yu, and D. Yu, “Neural ambisonic encoding for multi-speaker scenarios using a circular microphone array,” in IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2025.arxiv.org/abs/2409.06954
- [15]M. Heikkinen, A. Politis, K. Drossos, and T. Virtanen, “Gen-A: Generalizing ambisonics neural encoding to unseen microphone arrays,” in IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2025.arxiv.org/abs/2501.08047
- [16]“Beyond omnidirectional: Neural ambisonics encoding for arbitrary microphone directivity patterns using cross-attention,” in ICASSP 2026-2026 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP). IEEE, 2026, pp. 22 587–22 591.
- [17]A. Wabnitz, N. Epain, A. McEwan, and C. Jin, “Upscaling ambisonic sound scenes using compressed sensing techniques,” in IEEE Workshop on Applications of Signal Processing to Audio and Acoustics (WASPAA), 2011.
- [18]A. Wabnitz, N. Epain, and C. T. Jin, “A frequency-domain algorithm to upscale ambisonic sound scenes,” in IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2012, pp. 385–388.DOI
- [19]N. Epain and C. T. Jin, “Super-resolution sound field imaging with sub-space pre-processing,” in IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2013, pp. 350–354.DOI
- [20]G. Routray and R. M. Hegde, “Sparse plane-wave decomposition for upscaling ambisonic signals,” in IEEE International Conference on Signal Processing and Communications (SPCOM), 2020.DOI
- [21]M. Kentgens, S. Al Hares, and P. Jax, “On the upscaling of higher-order ambisonics signals for sound field translation,” in European Signal Processing Conference (EUSIPCO). IEEE, 2021, pp. 81–85.DOI
- [22]A. Politis, J. Vilkamo, and V. Pulkki, “Sector-based parametric sound field reproduction in the spherical harmonic domain,” vol. 9, no. 5, pp. 852–866, 2015.DOI
- [23]G. Routray, S. Basu, P. Baldev, and R. M. Hegde, “Deep-sound field analysis for upscaling ambisonic signals,” in EAA Spatial Audio Signal Processing Symposium, 2019.DOI
- [24]S. Gao, J. Lin, X. Wu, and T. Qu, “Sparse DNN model for frequency expanding of higher order ambisonics encoding process,” vol. 30, pp. 1124–1135, 2022.DOI
- [25]E. Chatzimoustafa and P. Jax, “Higher-order ambisonics upscaling using gated recurrent units,” in European Signal Processing Conference (EUSIPCO). IEEE, 2025, pp. 171–175.
- [26]Y. Luo and N. Mesgarani, “Conv-tasnet: Surpassing ideal time–frequency magnitude masking for speech separation,” vol. 27, no. 8, pp. 1256–1266, 2019.DOI
- [27]I. Nawfal et al., “Ambisonics super-resolution using a waveform-domain neural network,” in AES Conference on Audio for Virtual and Augmented Reality, 2024.arxiv.org/abs/2508.00240
- [28]F. Miotello, F. Terminiello, M. Pezzoli, A. Bernardini, F. Antonacci, and A. Sarti, “A physics-informed neural network-based approach for the spatial upsampling of spherical microphone arrays,” in IEEE International Workshop on Acoustic Signal Enhancement (IWAENC), 2024, pp. 215–219.arxiv.org/abs/2407.18732
- [29]L. Zhang, X. Wang, R. Hu, D. Li, and W. Tu, “Optimization of sound fields reproduction based higher-order ambisonics (HOA) using the generative adversarial network (GAN),” Multimedia Tools and Applications, vol. 80, no. 2, pp. 2205–2220, 2021.DOI
- [30]A. Milstein, N. Shlezinger, and B. Rafaely, “DiffAU: Diffusion-based ambisonics upscaling,” 2026.arxiv.org/abs/2510.00180
- [31]E. Picard, D. Di Carlo, A. A. Nugraha, M. Fontaine, and K. Yoshii, “SIRUP: A diffusion-based virtual unmixer of steering vectors for highly-directive spatialization with first-order ambisonics,” arXiv preprint arXiv:2602.17732, 2026.arxiv.org/abs/2602.17732
- [32]A. Milstein, N. Shlezinger, and B. Rafaely, “Generative learning for ambisonic upscaling,” 2026. [Online]. Available: https://arxiv.org/abs/2609.23479
- [33]H. Chung, J. Kim, M. T. McCann, M. L. Klasky, and J. C. Ye, “Diffusion posterior sampling for general noisy inverse problems,” in International Conference on Learning Representations (ICLR), 2022.arxiv.org/abs/2209.14687
- [34]E. Moliner, C. Hold, J. A. Ortiz, S. Prepelita, N. Ananthabhotla, D. Wong, S. Parekh, and S. Lee, “Ambisonics encoding of room impulse responses using a device-agnostic diffusion model,” arXiv preprint arXiv:2608.14097, 2026.arxiv.org/abs/2608.14097
- [35]B. Rafaely, Fundamentals of spherical array processing. Springer, 2015, vol. 8.
- [36]Y. Song, J. Sohl-Dickstein, D. P. Kingma, A. Kumar, S. Ermon, and B. Poole, “Score-based generative modeling through stochastic differential equations,” arXiv preprint arXiv:2011.13456, 2020.arxiv.org/abs/2011.13456
- [37]T. Karras, M. Aittala, T. Aila, and S. Laine, “Elucidating the design space of diffusion-based generative models,” Advances in neural information processing systems, vol. 35, pp. 26 565–26 577, 2022.arxiv.org/abs/2206.00364
- [38]H. E. Robbins, “An empirical bayes approach to statistics,” in Breakthroughs in Statistics: Foundations and basic theory. Springer, 1992, pp. 388–394.
- [39]Y. Song, J. Sohl-Dickstein, D. P. Kingma, A. Kumar, S. Ermon, and B. Poole, “Score-based generative modeling through stochastic differential equations,” arXiv preprint arXiv:2011.13456, 2020.arxiv.org/abs/2011.13456
- [40]E. Moliner, J. Lehtinen, and V. Välimäki, “Solving audio inverse problems with a diffusion model,” in IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2023.arxiv.org/abs/2210.15228
- [41]T. Gerkmann and R. Martin, “Empirical distributions of dft-domain speech coefficients based on estimated speech variances,” in Proc. Int. Workshop Acoust. Echo Noise Control, 2010.
- [42]J. Richter, S. Welker, J.-M. Lemercier, B. Lay, and T. Gerkmann, “Speech enhancement and dereverberation with diffusion-based generative models,” vol. 31, pp. 2351–2364, 2023.DOI
- [43]N. Liu, S. Li, Y. Du, A. Torralba, and J. B. Tenenbaum, “Compositional visual generation with composable diffusion models,” in European conference on computer vision. Springer, 2022, pp. 423–439.arxiv.org/abs/2206.01714
- [44]S. Kim, H. Kim, and S. Yoon, “Guided-tts 2: A diffusion model for high-quality adaptive text-to-speech with untranscribed data,” arXiv preprint arXiv:2205.15370, 2022.arxiv.org/abs/2205.15370
- [45]S. Saini and J. Peissig, “HARP: A large-scale higher-order ambisonic room impulse response dataset,” in IEEE International Conference on Acoustics, Speech, and Signal Processing Workshops (ICASSPW), 2025.arxiv.org/abs/2411.14207
- [46]J. Yamagishi, C. Veaux, and K. MacDonald, “CSTR VCTK Corpus,” 2019.
- [47]J. S. Garofolo, D. Graff, D. B. Paul, and D. S. Pallett, “CSR-I (WSJ0) complete,” https://catalog.ldc.upenn.edu/LDC93S6A, 1993.
- [48]J. Engel et al., “Project Aria: A new tool for egocentric multi-modal AI research,” arXiv preprint arXiv:2308.13561, 2023.arxiv.org/abs/2308.13561
- [49]J.-M. Lemercier, J. Richter, S. Welker, and T. Gerkmann, “Analyzing diffusion-based generative approaches versus discriminative approaches for speech restoration,” in IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2023.arxiv.org/abs/2211.02397
- [50]W. Peebles and S. Xie, “Scalable diffusion models with transformers,” in Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV), 2023, pp. 4195–4205.
- [51]“Method for the subjective assessment of intermediate quality level of audio systems,” International Telecommunication Union Radiocommunication Assembly, vol. 2, 2014.