Requirements Elicitation Follow-Up Question Generation

Interviews are a widely used technique in eliciting requirements to gather stakeholder needs, preferences, and expectations for a software system. Effective interviewing requires skilled interviewers to formulate appropriate interview questions in real time while facing multiple challenges, including lack of familiarity with the domain, excessive cognitive load, and information overload that hinders how humans process stakeholders' speech. Recently, large language models (LLMs) have exhibited state-of-the-art performance in multiple natural language processing tasks, including text summarization and entailment. To support interviewers, we investigate the application of GPT-4o to generate follow-up interview questions during requirements elicitation by building on a framework of common interviewer mistake types. In addition, we describe methods to generate questions based on interviewee speech. We report a controlled experiment to evaluate LLM-generated and human-authored questions with minimal guidance, and a second controlled experiment to evaluate the LLM-generated questions when generation is guided by interviewer mistake types. Our findings demonstrate that, for both experiments, the LLM-generated questions are no worse than the human-authored questions with respect to clarity, relevancy, and informativeness. In addition, LLM-generated questions outperform human-authored questions when guided by common mistakes types. This highlights the potential of using LLMs to help interviewers improve the quality and ease of requirements elicitation interviews in real time.

Self-Steering Deep Non-Linear Spatially Selective Filters for Efficient Extraction of Moving Speakers under Weak Guidance

Recent works on deep non-linear spatially selective filters demonstrate exceptional enhancement performance with computationally lightweight architectures for stationary speakers of known directions. However, to maintain this performance in dynamic scenarios, resource-intensive data-driven tracking algorithms become necessary to provide precise spatial guidance conditioned on the initial direction of a target speaker. As this additional computational overhead hinders application in resource-constrained scenarios such as real-time speech enhancement, we present a novel strategy utilizing a low-complexity tracking algorithm in the form of a particle filter instead. Assuming a causal, sequential processing style, we introduce temporal feedback to leverage the enhanced speech signal of the spatially selective filter to compensate for the limited modeling capabilities of the particle filter. Evaluation on a synthetic dataset illustrates how the autoregressive interplay between both algorithms drastically improves tracking accuracy and leads to strong enhancement performance. A listening test with real-world recordings complements these findings by indicating a clear trend towards our proposed self-steering pipeline as preferred choice over comparable methods.

DeSTA2.5-Audio: Toward General-Purpose Large Audio Language Model with Self-Generated Cross-Modal Alignment

We introduce DeSTA2.5-Audio, a general-purpose Large Audio Language Model (LALM) designed for robust auditory perception and instruction-following, without requiring task-specific audio instruction-tuning. Recent LALMs typically augment Large Language Models (LLMs) with auditory capabilities by training on large-scale, manually curated or LLM-synthesized audio-instruction datasets. However, these approaches have often suffered from the catastrophic forgetting of the LLM's original language abilities. To address this, we revisit the data construction pipeline and propose DeSTA, a self-generated cross-modal alignment strategy in which the backbone LLM generates its own training targets. This approach preserves the LLM's native language proficiency while establishing effective audio-text alignment, thereby enabling zero-shot generalization without task-specific tuning. Using DeSTA, we construct DeSTA-AQA5M, a large-scale, task-agnostic dataset containing 5 million training samples derived from 7,000 hours of audio spanning 50 diverse datasets, including speech, environmental sounds, and music. DeSTA2.5-Audio achieves state-of-the-art or competitive performance across a wide range of audio-language benchmarks, including Dynamic-SUPERB, MMAU, SAKURA, Speech-IFEval, and VoiceBench. Comprehensive comparative studies demonstrate that our self-generated strategy outperforms widely adopted data construction and training strategies in both auditory perception and instruction-following capabilities. Our findings underscore the importance of carefully designed data construction in LALM development and offer practical insights for building robust, general-purpose LALMs.

Measurement of the Granularity of Vowel Production Space By Just Producible Different (JPD) Limens

A body of work over the past several decades has demonstrated that the complex and coordinated articulatory movements of human vowel production are governed (at least in part)by control mechanisms whose targets are regions of auditory space. Within the target region control at the sub-phonemic level has also been demonstrated. But the degree of accuracy of that control is unknown. The current work investigates this question by asking how far apart must two vowel stimuli lie in auditory space in order to yield reliably different imitations? This distance is termed 'Just Producible Difference' (JPD). The current study uses a vowel mimicry paradigm to derive the first measurement of JPD among two sets of English speakers during front vowel production. JPD is estimated at between 14 and 51 mels in F1 X F2 space. This finding has implications for episodic theories of speech production. It also clarifies the possible structures of human vowel systems, by setting a theoretical lower bound for how close two vowel phonemes may be in a speaker's formant space, and hence a psychophysical explanation of observed trends in number and patterns of possible vowel phonemes.

De-AntiFake: Rethinking the Protective Perturbations Against Voice Cloning Attacks

The rapid advancement of speech generation models has heightened privacy and security concerns related to voice cloning (VC). Recent studies have investigated disrupting unauthorized voice cloning by introducing adversarial perturbations. However, determined attackers can mitigate these protective perturbations and successfully execute VC. In this study, we conduct the first systematic evaluation of these protective perturbations against VC under realistic threat models that include perturbation purification. Our findings reveal that while existing purification methods can neutralize a considerable portion of the protective perturbations, they still lead to distortions in the feature space of VC models, which degrades the performance of VC. From this perspective, we propose a novel two-stage purification method: (1) Purify the perturbed speech; (2) Refine it using phoneme guidance to align it with the clean speech distribution. Experimental results demonstrate that our method outperforms state-of-the-art purification methods in disrupting VC defenses. Our study reveals the limitations of adversarial perturbation-based VC defenses and underscores the urgent need for more robust solutions to mitigate the security and privacy risks posed by VC. The code and audio samples are available at https://de-antifake.github.io.

Multi-Utterance Speech Separation and Association Trained on Short Segments

Current deep neural network (DNN) based speech separation faces a fundamental challenge -- while the models need to be trained on short segments due to computational constraints, real-world applications typically require processing significantly longer recordings with multiple utterances per speaker than seen during training. In this paper, we investigate how existing approaches perform in this challenging scenario and propose a frequency-temporal recurrent neural network (FTRNN) that effectively bridges this gap. Our FTRNN employs a full-band module to model frequency dependencies within each time frame and a sub-band module that models temporal patterns in each frequency band. Despite being trained on short fixed-length segments of 10 s, our model demonstrates robust separation when processing signals significantly longer than training segments (21-121 s) and preserves speaker association across utterance gaps exceeding those seen during training. Unlike the conventional segment-separation-stitch paradigm, our lightweight approach (0.9 M parameters) performs inference on long audio without segmentation, eliminating segment boundary distortions while simplifying deployment. Experimental results demonstrate the generalization ability of FTRNN for multi-utterance speech separation and speaker association.

Open-Source System for Multilingual Translation and Cloned Speech Synthesis

We present an open-source system designed for multilingual translation and speech regeneration, addressing challenges in communication and accessibility across diverse linguistic contexts. The system integrates Whisper for speech recognition with Voice Activity Detection (VAD) to identify speaking intervals, followed by a pipeline of Large Language Models (LLMs). For multilingual applications, the first LLM segments speech into coherent, complete sentences, which a second LLM then translates. For speech regeneration, the system uses a text-to-speech (TTS) module with voice cloning capabilities to replicate the original speaker's voice, maintaining naturalness and speaker identity. The system's open-source components can operate locally or via APIs, offering cost-effective deployment across various use cases. These include real-time multilingual translation in Zoom sessions, speech regeneration for public broadcasts, and Bluetooth-enabled multilingual playback through personal devices. By preserving the speaker's voice, the system ensures a seamless and immersive experience, whether translating or regenerating speech. This open-source project is shared with the community to foster innovation and accessibility. We provide a detailed system performance analysis, including latency and word accuracy, demonstrating its potential to enable inclusive, adaptable communication solutions in real-world multilingual scenarios.

A Cookbook for Community-driven Data Collection of Impaired Speech in LowResource Languages

This study presents an approach for collecting speech samples to build Automatic Speech Recognition (ASR) models for impaired speech, particularly, low-resource languages. It aims to democratize ASR technology and data collection by developing a "cookbook" of best practices and training for community-driven data collection and ASR model building. As a proof-of-concept, this study curated the first open-source dataset of impaired speech in Akan: a widely spoken indigenous language in Ghana. The study involved participants from diverse backgrounds with speech impairments. The resulting dataset, along with the cookbook and open-source tools, are publicly available to enable researchers and practitioners to create inclusive ASR technologies tailored to the unique needs of speech impaired individuals. In addition, this study presents the initial results of fine-tuning open-source ASR models to better recognize impaired speech in Akan.

Benchmarking Akan ASR Models Across Domain-Specific Datasets: A Comparative Evaluation of Performance, Scalability, and Adaptability

Most existing automatic speech recognition (ASR) research evaluate models using in-domain datasets. However, they seldom evaluate how they generalize across diverse speech contexts. This study addresses this gap by benchmarking seven Akan ASR models built on transformer architectures, such as Whisper and Wav2Vec2, using four Akan speech corpora to determine their performance. These datasets encompass various domains, including culturally relevant image descriptions, informal conversations, biblical scripture readings, and spontaneous financial dialogues. A comparison of the word error rate and character error rate highlighted domain dependency, with models performing optimally only within their training domains while showing marked accuracy degradation in mismatched scenarios. This study also identified distinct error behaviors between the Whisper and Wav2Vec2 architectures. Whereas fine-tuned Whisper Akan models led to more fluent but potentially misleading transcription errors, Wav2Vec2 produced more obvious yet less interpretable outputs when encountering unfamiliar inputs. This trade-off between readability and transparency in ASR errors should be considered when selecting architectures for low-resource language (LRL) applications. These findings highlight the need for targeted domain adaptation techniques, adaptive routing strategies, and multilingual training frameworks for Akan and other LRLs.

Posterior Transition Modeling for Unsupervised Diffusion-Based Speech Enhancement

We explore unsupervised speech enhancement using diffusion models as expressive generative priors for clean speech. Existing approaches guide the reverse diffusion process using noisy speech through an approximate, noise-perturbed likelihood score, combined with the unconditional score via a trade-off hyperparameter. In this work, we propose two alternative algorithms that directly model the conditional reverse transition distribution of diffusion states. The first method integrates the diffusion prior with the observation model in a principled way, removing the need for hyperparameter tuning. The second defines a diffusion process over the noisy speech itself, yielding a fully tractable and exact likelihood score. Experiments on the WSJ0-QUT and VoiceBank-DEMAND datasets demonstrate improved enhancement metrics and greater robustness to domain shifts compared to both supervised and unsupervised baselines.