arXiv subjects

Beena Ahmed

Publications and source records attributed to Beena Ahmed.

At least 19 recordsLinked to original sources

arXiv2026

Mandarin Humorous Homophone Recognition and Disambiguation in Automatic Speech Recognition

Mandarin homophones remain a key challenge to improving automatic speech recognition (ASR) accuracy due to the amount of potential homophones. Mandarin speakers use this feature casually to convey emotions such as humour. Recent homophone-aware ASR studies have improved recognition accuracy, but intentional homophone twists in speech remain underexplored. In this paper, we identify patterns of homophone-based rhetorical wordplay in Mandarin, referred to as HumourPhone, and propose an ASR Adapter for homophone and HumourPhone recovery. Experimental results show that the proposed approach improves the recall of recognising HumourPhone by over 5\% and achieves a 4.35\% drop in target-span character error rate for homophone correction compared to baseline. These results highlight the need for homophone-aware modelling of lexical ambiguity and rhetorical wordplay in Mandarin speech recognition.

eess.AS↗

arXiv2026

Using Phonological-Level Wav2Vec2 for Mandarin Automatic Mispronunciation Detection and Diagnosis

Automatic mispronunciation detection and diagnosis (MDD) plays a crucial role in L2 Mandarin pronunciation learning. While end-to-end (E2E) based MDD methods have substantially improved phoneme-level detection accuracy, diagnostic feedback remains limited, as segmental and tonal errors are not explicitly separated. In this paper, we propose a phonological feature-based MDD framework that models both segmental and tonal attributes within a unified Wav2Vec2 CTC architecture. Experimental results show that the proposed method reduces the False Acceptance Rate (FAR) by 10.1% and the Diagnostic Error Rate (DER) by 23.6% compared with the phoneme-only baseline system. By decomposing phonemes into low-level phonological components, the proposed approach enables more detailed and interpretable diagnostic feedback for L2 learners.

eess.AS↗

arXiv2026

The WER Trap: Shattering the Illusion of Unified Tokens in Speech Language Models

The pursuit of a "unified" discrete token for both speech understanding and generation has led the Speech Language Model (SLM) community to heavily rely on Word Error Rate (WER) -- the core metric for Whisper-style tokenizers -- as the definitive proxy for representation quality. This fosters the assumption that low-WER tokens inherently preserve the information necessary for intelligible acoustic synthesis. We argue this is fundamentally deceptive. While high-frequency tokens succeed in generation tasks due to implicit information leakage, isolating pure semantic information at ultra-low frame rates strips away the finegrained articulation and micro-dynamics essential for ODE-based generation. Empirically validating this requires extreme compression without sacrificing WER -- a methodological bottleneck, as standard fixed-stride downsampling arbitrarily truncates phonetic boundaries. To overcome this, we develop a dynamic compression tokenizer that intelligently aligns representations with semantic boundaries, achieving ultra-low frame rates with exceptionally low WER. Using these isolated "pure" semantic tokens, we expose the WER trap: when conditioning generative models -- even with oracle duration alignments -- the reconstructed speech suffers from severe articulation blur and is rendered acoustically unintelligible. Our findings demonstrate that semantic categorization rewarded by low WER is inherently orthogonal to the continuous phonetic trajectories required for synthesis, shattering the illusion of the unified token and advocating for explicitly decoupled speech representations.

eess.AS↗

arXiv2026

Why Your Tokenizer Fails in Information Fusion: A Timing-Aware Pre-Quantization Fusion for Video-Enhanced Audio Tokenization

Audio tokenization has emerged as a critical component in end-to-end audio language models, enabling efficient discrete representation learning for both audio understanding and generation tasks. However, existing audio tokenizers face fundamental limitations in understanding tasks due to single-modality constraints, particularly when audio signals contain ambiguous or incomplete information. While incorporating additional modality information can significantly enhance audio understanding, current multimodal fusion approaches invariably degrade reconstruction quality. This degradation is unacceptable for end-to-end audio systems that require high-fidelity audio generation capabilities. In this work, we investigate the root causes of reconstruction quality degradation in video-enhanced audio tokenization and present three key findings. First, the location of fusion within the tokenizer architecture is crucial for preserving reconstruction quality. Second, we show that contrastive learning, though effective in continuous representation fusion, is unsuitable for discrete tokenizers as it fails to enhance downstream task performance. Third, while feature-dimension fusion approaches achieve moderate success, we discover that fusing along the temporal axis -- guided by the concept of distinctive features -- yields significantly better results. Building on these insights, we introduce the Timing-Aware Pre-Quantization Fusion for Video-Enhanced Audio Tokenization, the first approach to successfully integrate visual information into audio tokenizer architectures while preserving reconstruction fidelity. Our approach not only maintains high-fidelity reconstruction but also achieves superior performance on downstream understanding tasks compared with audio-only tokenizers and established multimodal fusion baselines.

eess.AS↗

arXiv2025

System X: A Mobile Voice-Based AI System for EMR Generation and Clinical Decision Support in Low-Resource Maternal Healthcare

We present the design, implementation, and in-situ deployment of a smartphone-based voice-enabled AI system for generating electronic medical records (EMRs) and clinical risk alerts in maternal healthcare settings. Targeted at low-resource environments such as Pakistan, the system integrates a fine-tuned, multilingual automatic speech recognition (ASR) model and a prompt-engineered large language model (LLM) to enable healthcare workers to engage naturally in Urdu, their native language, regardless of literacy or technical background. Through speech-based input and localized understanding, the system generates structured EMRs and flags critical maternal health risks. Over a seven-month deployment in a not-for-profit hospital, the system supported the creation of over 500 EMRs and flagged over 300 potential clinical risks. We evaluate the system's performance across speech recognition accuracy, EMR field-level correctness, and clinical relevance of AI-generated red flags. Our results demonstrate that speech based AI interfaces, can be effectively adapted to real-world healthcare settings, especially in low-resource settings, when combined with structured input design, contextual medical dictionaries, and clinician-in-the-loop feedback loops. We discuss generalizable design principles for deploying voice-based mobile healthcare AI support systems in linguistically and infrastructurally constrained settings.

cs.HC↗

arXiv2025

Zero-Shot Cognitive Impairment Detection from Speech Using AudioLLM

Cognitive impairment (CI) is of growing public health concern, and early detection is vital for effective intervention. Speech has gained attention as a non-invasive and easily collectible biomarker for assessing cognitive decline. Traditional CI detection methods typically rely on supervised models trained on acoustic and linguistic features extracted from speech, which often require manual annotation and may not generalise well across datasets and languages. In this work, we propose the first zero-shot speech-based CI detection method using the Qwen2- Audio AudioLLM, a model capable of processing both audio and text inputs. By designing prompt-based instructions, we guide the model in classifying speech samples as indicative of normal cognition or cognitive impairment. We evaluate our approach on two datasets: one in English and another multilingual, spanning different cognitive assessment tasks. Our results show that the zero-shot AudioLLM approach achieves performance comparable to supervised methods and exhibits promising generalizability and consistency across languages, tasks, and datasets.

cs.SD↗

arXiv2025

Distinctive Feature Codec: An Adaptive Efficient Speech Representation for Depression Detection

Large Language Models (LLMs) have demonstrated remarkable success across diverse fields, establishing a powerful paradigm for complex information processing. This has inspired the integration of speech into LLM frameworks, often by tokenizing continuous audio via neural speech codecs, enabling powerful speech language models. However, this dominant tokenization strategy relies on uniform frame-based processing at fixed time intervals. This fixed-rate approach, while effective for linguistic content, destroys the temporal dynamics. These dynamics are not noise but are established as primary biomarkers in clinical applications such as depression detection. To address this gap, we introduce the Distinctive Feature Codec (DFC), an adaptive framework engineered to preserve this vital timing information. Drawing from linguistic theory, DFC abandons fixed-interval processing and instead learns to dynamically segment the signal at perceptually significant acoustic transitions. This generates variable-length tokens that efficiently encode the temporal structure. As a key contribution, this work is the first to integrate traditional distinctive features into a modern deep learning codec for a temporally sensitive task such as depression detection. We also introduce the Group-wise Scalar Quantization (GSQ) approach to stably quantize these variable-length segments. Our distinctive feature-based approach offers a promising alternative to conventional frame-based processing and advances interpretable representation learning in the modern deep learning speech depression detection framework.

eess.AS↗

arXiv2025

Auto-Landmark: Acoustic Landmark Dataset and Open-Source Toolkit for Landmark Extraction

In the speech signal, acoustic landmarks identify times when the acoustic manifestations of the linguistically motivated distinctive features are most salient. Acoustic landmarks have been widely applied in various domains, including speech recognition, speech depression detection, clinical analysis of speech abnormalities, and the detection of disordered speech. However, there is currently no dataset available that provides precise timing information for landmarks, which has been proven to be crucial for downstream applications involving landmarks. In this paper, we selected the most useful acoustic landmarks based on previous research and annotated the TIMIT dataset with them, based on a combination of phoneme boundary information and manual inspection. Moreover, previous landmark extraction tools were not open source or benchmarked, so to address this, we developed an open source Python-based landmark extraction tool and established a series of landmark detection baselines. The first of their kinds, the dataset with landmark precise timing information, landmark extraction tool and baselines are designed to support a wide variety of future research.

eess.AS↗

arXiv2025

SpeechT-RAG: Reliable Depression Detection in LLMs with Retrieval-Augmented Generation Using Speech Timing Information

Large Language Models (LLMs) have been increasingly adopted for health-related tasks, yet their performance in depression detection remains limited when relying solely on text input. While Retrieval-Augmented Generation (RAG) typically enhances LLM capabilities, our experiments indicate that traditional text-based RAG systems struggle to significantly improve depression detection accuracy. This challenge stems partly from the rich depression-relevant information encoded in acoustic speech patterns information that current text-only approaches fail to capture effectively. To address this limitation, we conduct a systematic analysis of temporal speech patterns, comparing healthy individuals with those experiencing depression. Based on our findings, we introduce Speech Timing-based Retrieval-Augmented Generation, SpeechT-RAG, a novel system that leverages speech timing features for both accurate depression detection and reliable confidence estimation. This integrated approach not only outperforms traditional text-based RAG systems in detection accuracy but also enhances uncertainty quantification through a confidence scoring mechanism that naturally extends from the same temporal features. Our unified framework achieves comparable results to fine-tuned LLMs without additional training while simultaneously addressing the fundamental requirements for both accuracy and trustworthiness in mental health assessment.

eess.AS↗

arXiv2025

Mamba in Speech: Towards an Alternative to Self-Attention

Transformer and its derivatives have achieved success in diverse tasks across computer vision, natural language processing, and speech processing. To reduce the complexity of computations within the multi-head self-attention mechanism in Transformer, Selective State Space Models (i.e., Mamba) were proposed as an alternative. Mamba exhibited its effectiveness in natural language processing and computer vision tasks, but its superiority has rarely been investigated in speech signal processing. This paper explores solutions for applying Mamba to speech processing by discussing two typical speech processing tasks: speech recognition, which requires semantic and sequential information, and speech enhancement, which focuses primarily on sequential patterns. The experimental results confirm that bidirectional Mamba (BiMamba) consistently outperforms vanilla Mamba, highlighting the advantages of a bidirectional design for speech processing. Moreover, experiments demonstrate the effectiveness of BiMamba as an alternative to the self-attention module in the Transformer model and its derivates, particularly for the semantic-aware task. The crucial technologies for transferring Mamba to speech are then summarized in ablation studies and the discussion section, offering insights for extending this research to a broader scope of tasks.

eess.AS↗

arXiv2025

Why Pre-trained Models Fail: Feature Entanglement in Multi-modal Depression Detection

Depression remains a pressing global mental health issue, driving considerable research into AI-driven detection approaches. While pre-trained models, particularly speech self-supervised models (SSL Models), have been applied to depression detection, they show unexpectedly poor performance without extensive data augmentation. Large Language Models (LLMs), despite their success across various domains, have not been explored in multi-modal depression detection. In this paper, we first establish an LLM-based system to investigate its potential in this task, uncovering fundamental limitations in handling multi-modal information. Through systematic analysis, we discover that the poor performance of pre-trained models stems from the conflation of high-level information, where \textcolor{black}{one of the important reasons is that} high-level features derived from both content and speech are mixed within pre-trained models model representations, making it challenging to establish effective decision boundaries. To address this, we propose an information separation framework that disentangles these features, significantly improving the performance of both SSL models and LLMs in depression detection. Our experiments validate this finding and demonstrate that the integration of separated features yields substantial improvements over existing approaches, providing new insights for developing more effective multi-modal depression detection systems.

eess.AS↗

arXiv2024

When LLMs Meets Acoustic Landmarks: An Efficient Approach to Integrate Speech into Large Language Models for Depression Detection

Depression is a critical concern in global mental health, prompting extensive research into AI-based detection methods. Among various AI technologies, Large Language Models (LLMs) stand out for their versatility in mental healthcare applications. However, their primary limitation arises from their exclusive dependence on textual input, which constrains their overall capabilities. Furthermore, the utilization of LLMs in identifying and analyzing depressive states is still relatively untapped. In this paper, we present an innovative approach to integrating acoustic speech information into the LLMs framework for multimodal depression detection. We investigate an efficient method for depression detection by integrating speech signals into LLMs utilizing Acoustic Landmarks. By incorporating acoustic landmarks, which are specific to the pronunciation of spoken words, our method adds critical dimensions to text transcripts. This integration also provides insights into the unique speech patterns of individuals, revealing the potential mental states of individuals. Evaluations of the proposed approach on the DAIC-WOZ dataset reveal state-of-the-art results when compared with existing Audio-Text baselines. In addition, this approach is not only valuable for the detection of depression but also represents a new perspective in enhancing the ability of LLMs to comprehend and process speech signals.

eess.AS↗

arXiv2024

A Joint Spectro-Temporal Relational Thinking Based Acoustic Modeling Framework

Relational thinking refers to the inherent ability of humans to form mental impressions about relations between sensory signals and prior knowledge, and subsequently incorporate them into their model of their world. Despite the crucial role relational thinking plays in human understanding of speech, it has yet to be leveraged in any artificial speech recognition systems. Recently, there have been some attempts to correct this oversight, but these have been limited to coarse utterance-level models that operate exclusively in the time domain. In an attempt to narrow the gap between artificial systems and human abilities, this paper presents a novel spectro-temporal relational thinking based acoustic modeling framework. Specifically, it first generates numerous probabilistic graphs to model the relationships among speech segments across both time and frequency domains. The relational information rooted in every pair of nodes within these graphs is then aggregated and embedded into latent representations that can be utilized by downstream tasks. Models built upon this framework outperform state-of-the-art systems with a 7.82\% improvement in phoneme recognition tasks over the TIMIT dataset. In-depth analyses further reveal that our proposed relational thinking modeling mainly improves the model's ability to recognize vowels, which are the most likely to be confused by phoneme recognizers.

eess.AS↗

arXiv2024

Rethinking Mamba in Speech Processing by Self-Supervised Models

The Mamba-based model has demonstrated outstanding performance across tasks in computer vision, natural language processing, and speech processing. However, in the realm of speech processing, the Mamba-based model's performance varies across different tasks. For instance, in tasks such as speech enhancement and spectrum reconstruction, the Mamba model performs well when used independently. However, for tasks like speech recognition, additional modules are required to surpass the performance of attention-based models. We propose the hypothesis that the Mamba-based model excels in "reconstruction" tasks within speech processing. However, for "classification tasks" such as Speech Recognition, additional modules are necessary to accomplish the "reconstruction" step. To validate our hypothesis, we analyze the previous Mamba-based Speech Models from an information theory perspective. Furthermore, we leveraged the properties of HuBERT in our study. We trained a Mamba-based HuBERT model, and the mutual information patterns, along with the model's performance metrics, confirmed our assumptions.

eess.AS↗

arXiv2023

Variational Connectionist Temporal Classification for Order-Preserving Sequence Modeling

Connectionist temporal classification (CTC) is commonly adopted for sequence modeling tasks like speech recognition, where it is necessary to preserve order between the input and target sequences. However, CTC is only applied to deterministic sequence models, where the latent space is discontinuous and sparse, which in turn makes them less capable of handling data variability when compared to variational models. In this paper, we integrate CTC with a variational model and derive loss functions that can be used to train more generalizable sequence models that preserve order. Specifically, we derive two versions of the novel variational CTC based on two reasonable assumptions, the first being that the variational latent variables at each time step are conditionally independent; and the second being that these latent variables are Markovian. We show that both loss functions allow direct optimization of the variational lower bound for the model log-likelihood, and present computationally tractable forms for implementing them.

cs.LG↗

arXiv2023

Phonological Level wav2vec2-based Mispronunciation Detection and Diagnosis Method

The automatic identification and analysis of pronunciation errors, known as Mispronunciation Detection and Diagnosis (MDD) plays a crucial role in Computer Aided Pronunciation Learning (CAPL) tools such as Second-Language (L2) learning or speech therapy applications. Existing MDD methods relying on analysing phonemes can only detect categorical errors of phonemes that have an adequate amount of training data to be modelled. With the unpredictable nature of the pronunciation errors of non-native or disordered speakers and the scarcity of training datasets, it is unfeasible to model all types of mispronunciations. Moreover, phoneme-level MDD approaches have a limited ability to provide detailed diagnostic information about the error made. In this paper, we propose a low-level MDD approach based on the detection of speech attribute features. Speech attribute features break down phoneme production into elementary components that are directly related to the articulatory system leading to more formative feedback to the learner. We further propose a multi-label variant of the Connectionist Temporal Classification (CTC) approach to jointly model the non-mutually exclusive speech attributes using a single model. The pre-trained wav2vec2 model was employed as a core model for the speech attribute detector. The proposed method was applied to L2 speech corpora collected from English learners from different native languages. The proposed speech attribute MDD method was further compared to the traditional phoneme-level MDD and achieved a significantly lower False Acceptance Rate (FAR), False Rejection Rate (FRR), and Diagnostic Error Rate (DER) over all speech attributes compared to the phoneme-level equivalent.

eess.AS↗

arXiv2023

Spatial HuBERT: Self-supervised Spatial Speech Representation Learning for a Single Talker from Multi-channel Audio

Self-supervised learning has been used to leverage unlabelled data, improving accuracy and generalisation of speech systems through the training of representation models. While many recent works have sought to produce effective representations across a variety of acoustic domains, languages, modalities and even simultaneous speakers, these studies have all been limited to single-channel audio recordings. This paper presents Spatial HuBERT, a self-supervised speech representation model that learns both acoustic and spatial information pertaining to a single speaker in a potentially noisy environment by using multi-channel audio inputs. Spatial HuBERT learns representations that outperform state-of-the-art single-channel speech representations on a variety of spatial downstream tasks, particularly in reverberant and noisy environments. We also demonstrate the utility of the representations learned by Spatial HuBERT on a speech localisation downstream task. Along with this paper, we publicly release a new dataset of 100 000 simulated first-order ambisonics room impulse responses.

cs.CL↗

arXiv2022

Improving Children's Speech Recognition by Fine-tuning Self-supervised Adult Speech Representations

Children's speech recognition is a vital, yet largely overlooked domain when building inclusive speech technologies. The major challenge impeding progress in this domain is the lack of adequate child speech corpora; however, recent advances in self-supervised learning have created a new opportunity for overcoming this problem of data scarcity. In this paper, we leverage self-supervised adult speech representations and use three well-known child speech corpora to build models for children's speech recognition. We assess the performance of fine-tuning on both native and non-native children's speech, examine the effect of cross-domain child corpora, and investigate the minimum amount of child speech required to fine-tune a model which outperforms a state-of-the-art adult model. We also analyze speech recognition performance across children's ages. Our results demonstrate that fine-tuning with cross-domain child corpora leads to relative improvements of up to 46.08% and 45.53% for native and non-native child speech respectively, and absolute improvements of 14.70% and 31.10%. We also show that with as little as 5 hours of transcribed children's speech, it is possible to fine-tune a children's speech recognition system that outperforms a state-of-the-art adult model fine-tuned on 960 hours of adult speech.

cs.CL↗