TISDiSS: A Training-Time and Inference-Time Scalable Framework for Discriminative Source Separation

Source separation is a fundamental task in speech, music, and audio processing, and it also provides cleaner and larger data for training generative models. However, improving separation performance in practice often depends on increasingly large networks, inflating training and deployment costs. Motivated by recent advances in inference-time scaling for generative modeling, we propose Training-Time and Inference-Time Scalable Discriminative Source Separation (TISDiSS), a unified framework that integrates early-split multi-loss supervision, shared-parameter design, and dynamic inference repetitions. TISDiSS enables flexible speed-performance trade-offs by adjusting inference depth without retraining additional models. We further provide systematic analyses of architectural and training choices and show that training with more inference repetitions improves shallow-inference performance, benefiting low-latency applications. Experiments on standard speech separation benchmarks demonstrate state-of-the-art performance with a reduced parameter count, establishing TISDiSS as a scalable and practical framework for adaptive source separation.

ACMID: Automatic Curation of Musical Instrument Dataset for 7-Stem Music Source Separation

Most current music source separation (MSS) methods rely on supervised learning, limited by training data quan- tity and quality. Though web-crawling can bring abundant data, platform-level track labeling often causes metadata mismatches, impeding accurate "audio-label" pair acquisi- tion. To address this, we present ACMID: a dataset for MSS generated through web crawling of extensive raw data, fol- lowed by automatic cleaning via an instrument classifier built on a pre-trained audio encoder that filters and aggregates clean segments of target instruments from the crawled tracks, resulting in the refined ACMID-Cleaned dataset. Leverag- ing abundant data, we expand the conventional classifica- tion from 4-stem (Vocal/Bass/Drums/Others) to 7-stem (Pi- ano/Drums/Bass/Acoustic Guitar/Electric Guitar/Strings/Wind- Brass), enabling high granularity MSS systems. Experiments on SOTA MSS model demonstrates two key results: (i) MSS model trained with ACMID-Cleaned achieved a 2.39dB improvement in SDR performance compared to that with ACMID-Uncleaned, demostrating the effectiveness of our data cleaning procedure; (ii) incorporating ACMID-Cleaned to training enhances MSS model's average performance by 1.16dB, confirming the value of our dataset. Our data crawl- ing code, cleaning model code and weights are available at: https://github.com/scottishfold0621/ACMID.

Human Motion Video Generation: A Survey

Human motion video generation has garnered significant research interest due to its broad applications, enabling innovations such as photorealistic singing heads or dynamic avatars that seamlessly dance to music. However, existing surveys in this field focus on individual methods, lacking a comprehensive overview of the entire generative process. This paper addresses this gap by providing an in-depth survey of human motion video generation, encompassing over ten sub-tasks, and detailing the five key phases of the generation process: input, motion planning, motion video generation, refinement, and output. Notably, this is the first survey that discusses the potential of large language models in enhancing human motion video generation. Our survey reviews the latest developments and technological trends in human motion video generation across three primary modalities: vision, text, and audio. By covering over two hundred papers, we offer a thorough overview of the field and highlight milestone works that have driven significant technological breakthroughs. Our goal for this survey is to unveil the prospects of human motion video generation and serve as a valuable resource for advancing the comprehensive applications of digital humans. A complete list of the models examined in this survey is available in Our Repository https://github.com/Winn1y/Awesome-Human-Motion-Video-Generation.

MelCap: A Unified Single-Codebook Neural Codec for High-Fidelity Audio Compression

Neural audio codecs have recently emerged as powerful tools for high-quality and low-bitrate audio compression, leveraging deep generative models to learn latent representations of audio signals. However, existing approaches either rely on a single quantizer that only processes speech domain, or on multiple quantizers that are not well suited for downstream tasks. To address this issue, we propose MelCap, a unified "one-codebook-for-all" neural codec that effectively handles speech, music, and general sound. By decomposing audio reconstruction into two stages, our method preserves more acoustic details than previous single-codebook approaches, while achieving performance comparable to mainstream multi-codebook methods. In the first stage, audio is transformed into mel-spectrograms, which are compressed and quantized into compact single tokens using a 2D tokenizer. A perceptual loss is further applied to mitigate the over-smoothing artifacts observed in spectrogram reconstruction. In the second stage, a Vocoder recovers waveforms from the mel discrete tokens in a single forward pass, enabling real-time decoding. Both objective and subjective evaluations demonstrate that MelCap achieves quality on comparable to state-of-the-art multi-codebook codecs, while retaining the computational simplicity of a single-codebook design, thereby providing an effective representation for downstream tasks.

Improving Perceptual Audio Aesthetic Assessment via Triplet Loss and Self-Supervised Embeddings

We present a system for automatic multi-axis perceptual quality prediction of generative audio, developed for Track 2 of the AudioMOS Challenge 2025. The task is to predict four Audio Aesthetic Scores--Production Quality, Production Complexity, Content Enjoyment, and Content Usefulness--for audio generated by text-to-speech (TTS), text-to-audio (TTA), and text-to-music (TTM) systems. A main challenge is the domain shift between natural training data and synthetic evaluation data. To address this, we combine BEATs, a pretrained transformer-based audio representation model, with a multi-branch long short-term memory (LSTM) predictor and use a triplet loss with buffer-based sampling to structure the embedding space by perceptual similarity. Our results show that this improves embedding discriminability and generalization, enabling domain-robust audio quality assessment without synthetic training data.

Timbre-Adaptive Transcription: A Lightweight Architecture with Associative Memory for Dynamic Instrument Separation

Existing multi-timbre transcription models struggle with generalization beyond pre-trained instruments and rigid source-count constraints. We address these limitations with a lightweight deep clustering solution featuring: 1) a timbre-agnostic backbone achieving state-of-the-art performance with only half the parameters of comparable models, and 2) a novel associative memory mechanism that mimics human auditory cognition to dynamically encode unseen timbres via attention-based clustering. Our biologically-inspired framework enables adaptive polyphonic separation with minimal training data (12.5 minutes), supported by a new synthetic dataset method offering cost-effective, high-precision multi-timbre generation. Experiments show the timbre-agnostic transcription model outperforms existing models on public benchmarks, while the separation module demonstrates promising timbre discrimination. This work provides an efficient framework for timbre-related music transcription and explores new directions for timbre-aware separation through cognitive-inspired architectures.

MAIA: An Inpainting-Based Approach for Music Adversarial Attacks

Music adversarial attacks have garnered significant interest in the field of Music Information Retrieval (MIR). In this paper, we present Music Adversarial Inpainting Attack (MAIA), a novel adversarial attack framework that supports both white-box and black-box attack scenarios. MAIA begins with an importance analysis to identify critical audio segments, which are then targeted for modification. Utilizing generative inpainting models, these segments are reconstructed with guidance from the output of the attacked model, ensuring subtle and effective adversarial perturbations. We evaluate MAIA on multiple MIR tasks, demonstrating high attack success rates in both white-box and black-box settings while maintaining minimal perceptual distortion. Additionally, subjective listening tests confirm the high audio fidelity of the adversarial samples. Our findings highlight vulnerabilities in current MIR systems and emphasize the need for more robust and secure models.

Exploring Situated Stabilities of a Rhythm Generation System through Variational Cross-Examination

This paper investigates GrooveTransformer, a real-time rhythm generation system, through the postphenomenological framework of Variational Cross-Examination (VCE). By reflecting on its deployment across three distinct artistic contexts, we identify three stabilities: an autonomous drum accompaniment generator, a rhythmic control voltage sequencer in Eurorack format, and a rhythm driver for a harmonic accompaniment system. The versatility of its applications was not an explicit goal from the outset of the project. Thus, we ask: how did this multistability emerge? Through VCE, we identify three key contributors to its emergence: the affordances of system invariants, the interdisciplinary collaboration, and the situated nature of its development. We conclude by reflecting on the viability of VCE as a descriptive and analytical method for Digital Musical Instrument (DMI) design, emphasizing its value in uncovering how technologies mediate, co-shape, and are co-shaped by users and contexts.

VoiceAssistant-Eval: Benchmarking AI Assistants across Listening, Speaking, and Viewing

The growing capabilities of large language models and multimodal systems have spurred interest in voice-first AI assistants, yet existing benchmarks are inadequate for evaluating the full range of these systems' capabilities. We introduce VoiceAssistant-Eval, a comprehensive benchmark designed to assess AI assistants across listening, speaking, and viewing. VoiceAssistant-Eval comprises 10,497 curated examples spanning 13 task categories. These tasks include natural sounds, music, and spoken dialogue for listening; multi-turn dialogue, role-play imitation, and various scenarios for speaking; and highly heterogeneous images for viewing. To demonstrate its utility, we evaluate 21 open-source models and GPT-4o-Audio, measuring the quality of the response content and speech, as well as their consistency. The results reveal three key findings: (1) proprietary models do not universally outperform open-source models; (2) most models excel at speaking tasks but lag in audio understanding; and (3) well-designed smaller models can rival much larger ones. Notably, the mid-sized Step-Audio-2-mini (7B) achieves more than double the listening accuracy of LLaMA-Omni2-32B-Bilingual. However, challenges remain: multimodal (audio plus visual) input and role-play voice imitation tasks are difficult for current models, and significant gaps persist in robustness and safety alignment. VoiceAssistant-Eval identifies these gaps and establishes a rigorous framework for evaluating and guiding the development of next-generation AI assistants. Code and data will be released at https://mathllm.github.io/VoiceAssistantEval/ .

MAPSS: Manifold-based Assessment of Perceptual Source Separation

Objective assessment of source-separation systems still mismatches subjective human perception, especially when leakage and self-distortion interact. We introduce the Perceptual Separation (PS) and Perceptual Match (PM), the first pair of measures that functionally isolate these two factors. Our intrusive method begins with generating a bank of fundamental distortions for each reference waveform signal in the mixture. Distortions, references, and their respective system outputs from all sources are then independently encoded by a pre-trained self-supervised learning model. These representations are aggregated and projected onto a manifold via diffusion maps, which aligns Euclidean distances on the manifold with dissimilarities of the encoded waveforms. On this manifold, the PM measures the Mahalanobis distance from each output to its attributed cluster that consists of its reference and distortions embeddings, capturing self-distortion. The PS accounts for the Mahalanobis distance of the output to the attributed and to the closest non-attributed clusters, quantifying leakage. Both measures are differentiable and granular, operating at a resolution as low as 50 frames per second. We further derive, for both measures, deterministic error radius and non-asymptotic, high-probability confidence intervals (CIs). Experiments on English, Spanish, and music mixtures show that the PS and PM nearly always achieve the highest linear correlation coefficients with human mean-opinion scores than 14 competitors, reaching as high as 86.36% for speech and 87.21% for music. We observe, at worst, an error radius of 1.39% and a probabilistic 95% CI of 12.21% for these coefficients, which improves reliable and informed evaluation. Using mutual information, the measures complement each other most as their values decrease, suggesting they are jointly more informative as system performance degrades.