arXiv subjects

Jagabandhu Mishra

Publications and source records attributed to Jagabandhu Mishra.

At least 19 recordsLinked to original sources

arXiv2026

Improving ASR Fairness for Cleft Lip and Palate Speech: A Study on Severity-Aware Data Mixing

Speech produced by individuals with cleft lip and palate (CLP) is often hypernasal (and sometimes breathy) due to structural anomalies, yielding shifts in formant structure that degrade automatic speech recognition (ASR) performance and fairness. Building on evidence that mainstream ASR systems underperform on atypical and disordered speech, we posit that widely used services (e.g., Google Speech-to-Text) exhibit reduced fairness for CLP speech, and we evaluate this claim empirically. To quantify fairness consistently, we introduce a simple fairness score (FS) that trades off overall error and between-group disparity. Despite formant disruptions, mild and moderate CLP speech retains partial spectro-temporal alignment with typical speech, motivating the use of mixing strategies to improve fairness. We systematically investigated severity-aware mixing of CLP and normal speech at different severity levels and assessed its effect on fairness. Three ASR models GMM-HMM, Whisper, and XLSR were evaluated on the AIISH (Kannada language) and NMCPC (English language) datasets. A mixing strategy that leverages severity-aware mixing of CLP and normal speech improves fairness on both English (NMCPC) and Kannada (AIISH) corpora. Notably, the word error rate (WER) decreased from 37.58% to 25.47% (GMM-HMM, AIISH) and from 35.74% to 21.72% (Whisper, NMCPC).

eess.AS↗

arXiv2026

Normal-Anchored First-Order Model-Agnostic Meta-Learning based Whisper Fine-Tuning for Enhancing Fairness of Cleft Lip and Palate Speech Recognition

Automatic speech recognition (ASR) for cleft lip and palate (CLP) speech is difficult because acoustic and articulatory patterns vary across severity levels. This variability reduces the performance of pretrained ASR systems, and conventional fine-tuning may not generalize well under low-resource, heterogeneous CLP conditions. This work proposes Normal-Anchored First-Order Model-Agnostic Meta-Learning (NA-FOMAML) for adapting Whisper to CLP speech. The method uses a first-order bilevel meta-learning framework in which normal speech is used in the inner loop as a stable support condition, while CLP severity groups are used in the outer loop to improve post-adaptation robustness. This design aims to reduce the performance gap between normal and pathological speech. Experiments are conducted on the NMCPC and AIISH datasets using four normal-anchored training configurations. Frozen encoder, full encoder, and selected Whisper encoder-layer tuning strategies are evaluated, including layers 0--5, 4--11, 6--11, and 8--11, with decoder and projection-head adaptation. Results show that outer-loop training with only normal speech is insufficient. For NMCPC, full encoder tuning with Normal to Normal+Mild+Moderate gives WERs of 4.40%, 5.53%, 16.14%, and 52.07% for normal, mild, moderate, and severe speech. For AIISH, full encoder tuning with Normal to Normal+Mild+Moderate+Severe gives WERs of 2.48%, 19.66%, 14.05%, and 57.50%. A transcription-based phoneme-category analysis shows that severe CLP speech has high error rates across fricatives, affricates, nasals, liquids, plosives, and vowels. Overall, NA-FOMAML improves cross-severity robustness, but severe speech still requires severity-aware sampling, phoneme-aware loss functions, and augmentation targeting pressure consonant and resonance-related distortions.

eess.SP↗

arXiv2026

Phonetically Explainable Speech Deepfake Detection

Speech deepfake detection is predominantly treated as an opaque classification task where all temporal frames are aggregated equally. This ignores that different phonetic categories carry vastly different amounts of discriminative information. To address this, we propose a phoneme-guided cross-attention framework that transforms detection into an interpretable, phonetically grounded process. We factorize the spoofing posterior $P(\text{spoofed}\mid X, W)$, conditioned on the acoustic representation $X$ and the phonetic posteriorgram $W$. The resulting factorization can be written as $P(\text{spoofed} \mid X, W) = \sum_{i=1}^{M} w_i \cdot P(\text{spoofed} \mid X, Z = z_i)$, where $M$ denotes the number of phonetic classes, $P(\text{spoofed} \mid X, Z = z_i)$ is the spoofing probability for the $i$-th phonetic class $z_i$ conditioned on $X$, and each $w_i$ is the prevalence of phonetic class $z_i$ in the utterance. Our transformer-based architecture instantiates this through a cross-attention block in which phonetic queries selectively probe information in acoustic keys and values, with softmax-normalized pooling supplying explicit phone-presence weights. Unlike prior approaches that rely heavily on post-hoc explainability methods, our framework offers phonetic-explainability-by-design. We evaluate the framework on an LJSpeech-derived corpus, ASVspoof 2019 LA, and ASVspoof 5 Track 1. Per-phone importance rankings reveal that discriminative power concentrates on articulatory categories that generative models struggle to reproduce faithfully. Stops, fricatives, affricates, nasals, and silence-boundary closures rank most discriminative, while periodic vowels and semivowels rank lower. Beyond competitive performance, our model provides structural interpretability, yielding an inspectable per-articulatory category breakdown of the final verdict.

eess.AS↗

arXiv2026

Kinship Verification Using Voice

Kinship verification (KV) from voice, the task of determining whether two speakers are biologically related, has received only little attention. Our work establishes a foundational basis for this emerging frontier, contributing to both performance evaluation and detection methodologies. First, leveraging the speech recordings of the large-scale audio-visual dataset, KAN-AV, we propose a revised evaluation protocol that controls for various confounders and adopts a family-disjoint train--test split to address open-set KV. Second, we analyze the close connection between speaker verification and KV, showing that genealogical similarity of speaker pairs plays opposite roles in the two tasks. Third, we tackle KV using three neural speaker embedding extractors (ECAPA-TDNN, WavLM-ECAPA, and ReDimNet) combined with various back-ends. In zero-shot KV including same-speaker target trials, ReDimNet achieves the lowest equal error rate (EER) of $20.8\%$; however, performance degrades to $39.7\%$ under strict kin trials, where same-speaker target trials are excluded. Our best trainable back-end, which applies asymmetric processing of the embedding pair to mitigate age-difference effects, obtains an EER of $32.0\%$ ($18.6\%$ with speaker target trials included). These results highlight the difficulty of KV while showing that speaker embeddings encode familial cues, offering a promising foundation for voice-based kinship analysis.

eess.AS↗

arXiv2026

Advancing Zero-Shot Open-Set Speech Deepfake Source Tracing

We propose a novel zero-shot source tracing framework inspired by speaker verification. We adapt SSL-AASIST for attack classification, enhancing embeddings with AAM loss and RegMixup, and ensure that training attacks are disjoint from those forming fingerprint-trial pairs. For backend scoring in attack verification, we explore both zero-shot approaches (cosine similarity and Siamese) and few-shot approaches (MLP and Siamese). Experiments on our recently introduced STOPA dataset with an open set setting show that few-shot learning provides advantages in the in-distribution (ID) scenario, while zero-shot approaches perform better in the out-of-distribution (OOD) scenario. In attack source verification with ID trials, few-shot Siamese and MLP achieve equal error rates (EER) of 17.72% and 13.11%, compared to 29.91% for zero-shot cosine scoring. Conversely, in OOD trials, zero-shot cosine scoring reaches 16.43%, outperforming few-shot Siamese at 23.47% and MLP at 21.57%.

eess.AS↗

arXiv2026

Joint Optimization of Speaker and Spoof Detectors for Spoofing-Robust Automatic Speaker Verification

Spoofing-robust speaker verification (SASV) combines the tasks of speaker and spoof detection to authenticate speakers under adversarial settings. Many SASV systems rely on fusion of speaker and spoof cues at embedding, score or decision levels, based on independently trained subsystems. In this study, we respect similar modularity of the two subsystems, by integrating their outputs using trainable back-end classifiers. In particular, we explore various approaches for directly optimizing the back-end for the recently-proposed SASV performance metric (a-DCF) as a training objective. Our experiments on the ASVspoof 5 dataset demonstrate two important findings: (i) nonlinear score fusion consistently improves a-DCF over linear fusion, and (ii) the combination of weighted cosine scoring for speaker detection with SSL-AASIST for spoof detection achieves state-of-the-art performance, reducing min a-DCF to 0.196 and SPF-EER to 7.6%. These contributions highlight the importance of modular design, calibrated integration, and task-aligned optimization for advancing robust and interpretable SASV systems.

eess.AS↗

arXiv2026

Joint Optimization of ASV and CM tasks: BTUEF Team's Submission for WildSpoof Challenge

Spoofing-aware speaker verification (SASV) jointly addresses automatic speaker verification and spoofing countermeasures to improve robustness against adversarial attacks. In this paper, we investigate our recently proposed modular SASV framework that enables effective reuse of publicly available ASV and CM systems through non-linear fusion, explicitly modeling their interaction, and optimization with an operating-condition-dependent trainable a-DCF loss. The framework is evaluated using ECAPA-TDNN and ReDimNet as ASV embedding extractors and SSL-AASIST as the CM model, with experiments conducted both with and without fine-tuning on the WildSpoof SASV training data. Results show that the best performance is achieved by combining ReDimNet-based ASV embeddings with fine-tuned SSL-AASIST representations, yielding an a-DCF of 0.0515 on the progress evaluation set and 0.2163 on the final evaluation set.

eess.AS↗

arXiv2026

Towards Fair ASR For Second Language Speakers Using Fairness Prompted Finetuning

In this work, we address the challenge of building fair English ASR systems for second-language speakers. Our analysis of widely used ASR models, Whisper and Seamless-M4T, reveals large fluctuations in word error rate (WER) across 26 accent groups, indicating significant fairness gaps. To mitigate this, we propose fairness-prompted finetuning with lightweight adapters, incorporating Spectral Decoupling (SD), Group Distributionally Robust Optimization (Group-DRO), and Invariant Risk Minimization (IRM). Our proposed fusion of traditional empirical risk minimization (ERM) with cross-entropy and fairness-driven objectives (SD, Group DRO, and IRM) enhances fairness across accent groups while maintaining overall recognition accuracy. In terms of macro-averaged word error rate, our approach achieves a relative improvement of 58.7% and 58.5% over the large pretrained Whisper and SeamlessM4T, and 9.7% and 7.8% over them, finetuning with standard empirical risk minimization with cross-entropy loss.

cs.CL↗

arXiv2025

STOPA: A Database of Systematic VariaTion Of DeePfake Audio for Open-Set Source Tracing and Attribution

A key research area in deepfake speech detection is source tracing - determining the origin of synthesised utterances. The approaches may involve identifying the acoustic model (AM), vocoder model (VM), or other generation-specific parameters. However, progress is limited by the lack of a dedicated, systematically curated dataset. To address this, we introduce STOPA, a systematically varied and metadata-rich dataset for deepfake speech source tracing, covering 8 AMs, 6 VMs, and diverse parameter settings across 700k samples from 13 distinct synthesisers. Unlike existing datasets, which often feature limited variation or sparse metadata, STOPA provides a systematically controlled framework covering a broader range of generative factors, such as the choice of the vocoder model, acoustic model, or pretrained weights, ensuring higher attribution reliability. This control improves attribution accuracy, aiding forensic analysis, deepfake detection, and generative model transparency.

cs.SD↗

arXiv2025

Fusion of Modulation Spectrogram and SSL with Multi-head Attention for Fake Speech Detection

Fake speech detection systems have become a necessity to combat against speech deepfakes. Current systems exhibit poor generalizability on out-of-domain speech samples due to lack to diverse training data. In this paper, we attempt to address domain generalization issue by proposing a novel speech representation using self-supervised (SSL) speech embeddings and the Modulation Spectrogram (MS) feature. A fusion strategy is used to combine both speech representations to introduce a new front-end for the classification task. The proposed SSL+MS fusion representation is passed to the AASIST back-end network. Experiments are conducted on monolingual and multilingual fake speech datasets to evaluate the efficacy of the proposed model architecture in cross-dataset and multilingual cases. The proposed model achieves a relative performance improvement of 37% and 20% on the ASVspoof 2019 and MLAAD datasets, respectively, in in-domain settings compared to the baseline. In the out-of-domain scenario, the model trained on ASVspoof 2019 shows a 36% relative improvement when evaluated on the MLAAD dataset. Across all evaluated languages, the proposed model consistently outperforms the baseline, indicating enhanced domain generalization.

eess.AS↗

arXiv2025

Parameter-Efficient Fine-Tuning of Foundation Models for CLP Speech Classification

We propose the use of parameter-efficient fine-tuning (PEFT) of foundation models for cleft lip and palate (CLP) detection and severity classification. In CLP, nasalization increases with severity due to the abnormal passage between the oral and nasal tracts; this causes oral stops to be replaced by glottal stops and alters formant trajectories and vowel space. Since foundation models are trained for grapheme prediction or long-term quantized representation prediction, they may better discriminate CLP severity when fine-tuned on domain-specific data. We conduct experiments on two datasets: English (NMCPC) and Kannada (AIISH). We perform a comparative analysis using embeddings from self-supervised models Wav2Vec2 and WavLM, and the weakly supervised Whisper, each paired with SVM classifiers, and compare them with traditional handcrafted features eGeMAPS and ComParE. Finally, we fine-tune the best-performing Whisper model using PEFT techniques: Low-Rank Adapter (LoRA) and Decomposed Low-Rank Adapter (DoRA). Our results demonstrate that the proposed approach achieves relative improvements of 26.4% and 63.4% in macro-average F1 score over the best foundation model and handcrafted feature baselines on the NMCPC dataset, and improvements of 6.1% and 52.9% on the AIISH dataset, respectively.

eess.AS↗

arXiv2025

Leveraging AM and FM Rhythm Spectrograms for Dementia Classification and Assessment

This study explores the potential of Rhythm Formant Analysis (RFA) to capture long-term temporal modulations in dementia speech. Specifically, we introduce RFA-derived rhythm spectrograms as novel features for dementia classification and regression tasks. We propose two methodologies: (1) handcrafted features derived from rhythm spectrograms, and (2) a data-driven fusion approach, integrating proposed RFA-derived rhythm spectrograms with vision transformer (ViT) for acoustic representations along with BERT-based linguistic embeddings. We compare these with existing features. Notably, our handcrafted features outperform eGeMAPs with a relative improvement of $14.2\%$ in classification accuracy and comparable performance in the regression task. The fusion approach also shows improvement, with RFA spectrograms surpassing Mel spectrograms in classification by around a relative improvement of $13.1\%$ and a comparable regression score with the baselines.

eess.AS↗

arXiv2025

Towards Explainable Spoofed Speech Attribution and Detection:a Probabilistic Approach for Characterizing Speech Synthesizer Components

We propose an explainable probabilistic framework for characterizing spoofed speech by decomposing it into probabilistic attribute embeddings. Unlike raw high-dimensional countermeasure embeddings, which lack interpretability, the proposed probabilistic attribute embeddings aim to detect specific speech synthesizer components, represented through high-level attributes and their corresponding values. We use these probabilistic embeddings with four classifier back-ends to address two downstream tasks: spoofing detection and spoofing attack attribution. The former is the well-known bonafide-spoof detection task, whereas the latter seeks to identify the source method (generator) of a spoofed utterance. We additionally use Shapley values, a widely used technique in machine learning, to quantify the relative contribution of each attribute value to the decision-making process in each task. Results on the ASVspoof2019 dataset demonstrate the substantial role of duration and conversion modeling in spoofing detection; and waveform generation and speaker modeling in spoofing attack attribution. In the detection task, the probabilistic attribute embeddings achieve $99.7\%$ balanced accuracy and $0.22\%$ equal error rate (EER), closely matching the performance of raw embeddings ($99.9\%$ balanced accuracy and $0.22\%$ EER). Similarly, in the attribution task, our embeddings achieve $90.23\%$ balanced accuracy and $2.07\%$ EER, compared to $90.16\%$ and $2.11\%$ with raw embeddings. These results demonstrate that the proposed framework is both inherently explainable by design and capable of achieving performance comparable to raw CM embeddings.

eess.AS↗

arXiv2025

Optimizing a-DCF for Spoofing-Robust Speaker Verification

Automatic speaker verification (ASV) systems are vulnerable to spoofing attacks. We propose a spoofing-robust ASV system optimized directly for the recently introduced architecture-agnostic detection cost function (a-DCF), which allows targeting a desired trade-off between the contradicting aims of user convenience and robustness to spoofing. We combine a-DCF and binary cross-entropy (BCE) with a novel straightforward threshold optimization technique. Our results with an embedding fusion system on ASVspoof2019 data demonstrate relative improvement of $13\%$ over a system trained using BCE only (from minimum a-DCF of $0.1445$ to $0.1254$). Using an alternative non-linear score fusion approach provides relative improvement of $43\%$ (from minimum a-DCF of $0.0508$ to $0.0289$).

eess.AS↗

arXiv2024

An Explainable Probabilistic Attribute Embedding Approach for Spoofed Speech Characterization

We propose a novel approach for spoofed speech characterization through explainable probabilistic attribute embeddings. In contrast to high-dimensional raw embeddings extracted from a spoofing countermeasure (CM) whose dimensions are not easy to interpret, the probabilistic attributes are designed to gauge the presence or absence of sub-components that make up a specific spoofing attack. These attributes are then applied to two downstream tasks: spoofing detection and attack attribution. To enforce interpretability also to the back-end, we adopt a decision tree classifier. Our experiments on the ASVspoof2019 dataset with spoof CM embeddings extracted from three models (AASIST, Rawboost-AASIST, SSL-AASIST) suggest that the performance of the attribute embeddings are on par with the original raw spoof CM embeddings for both tasks. The best performance achieved with the proposed approach for spoofing detection and attack attribution, in terms of accuracy, is 99.7% and 99.2%, respectively, compared to 99.7% and 94.7% using the raw CM embeddings. To analyze the relative contribution of each attribute, we estimate their Shapley values. Attributes related to acoustic feature prediction, waveform generation (vocoder), and speaker modeling are found important for spoofing detection; while duration modeling, vocoder, and input type play a role in spoofing attack attribution.

eess.AS↗

arXiv2024

Spoofing-Robust Speaker Verification Using Parallel Embedding Fusion: BTU Speech Group's Approach for ASVspoof5 Challenge

This paper introduces the parallel network-based spoofing-aware speaker verification (SASV) system developed by BTU Speech Group for the ASVspoof5 Challenge. The SASV system integrates ASV and CM systems to enhance security against spoofing attacks. Our approach employs score and embedding fusion from ASV models (ECAPA-TDNN, WavLM) and CM models (AASIST). The fused embeddings are processed using a simple DNN structure, optimizing model performance with a combination of recently proposed a-DCF and BCE losses. We introduce a novel parallel network structure where two identical DNNs, fed with different inputs, independently process embeddings and produce SASV scores. The final SASV probability is derived by averaging these scores, enhancing robustness and accuracy. Experimental results demonstrate that the proposed parallel DNN structure outperforms traditional single DNN methods, offering a more reliable and secure speaker verification system against spoofing attacks.

eess.AS↗

arXiv2023

Language vs Speaker Change: A Comparative Study

Spoken language change detection (LCD) refers to detecting language switching points in a multilingual speech signal. Speaker change detection (SCD) refers to locating the speaker change points in a multispeaker speech signal. The objective of this work is to understand the challenges in LCD task by comparing it with SCD task. Human subjective study for change detection is performed for LCD and SCD. This study demonstrates that LCD requires larger duration spectro-temporal information around the change point compared to SCD. Based on this, the work explores automatic distance based and model based LCD approaches. The model based ones include Gaussian mixture model and universal background model (GMM-UBM), attention, and Generative adversarial network (GAN) based approaches. Both the human and automatic LCD tasks infer that the performance of the LCD task improves by incorporating more and more spectro-temporal duration.

eess.AS↗

arXiv2023

Implicit Self-supervised Language Representation for Spoken Language Diarization

In a code-switched (CS) scenario, the use of spoken language diarization (LD) as a pre-possessing system is essential. Further, the use of implicit frameworks is preferable over the explicit framework, as it can be easily adapted to deal with low/zero resource languages. Inspired by speaker diarization (SD) literature, three frameworks based on (1) fixed segmentation, (2) change point-based segmentation and (3) E2E are proposed to perform LD. The initial exploration with synthetic TTSF-LD dataset shows, using x-vector as implicit language representation with appropriate analysis window length ($N$) can able to achieve at per performance with explicit LD. The best implicit LD performance of $6.38$ in terms of Jaccard error rate (JER) is achieved by using the E2E framework. However, considering the E2E framework the performance of implicit LD degrades to $60.4$ while using with practical Microsoft CS (MSCS) dataset. The difference in performance is mostly due to the distributional difference between the monolingual segment duration of secondary language in the MSCS and TTSF-LD datasets. Moreover, to avoid segment smoothing, the smaller duration of the monolingual segment suggests the use of a small value of $N$. At the same time with small $N$, the x-vector representation is unable to capture the required language discrimination due to the acoustic similarity, as the same speaker is speaking both languages. Therefore, to resolve the issue a self-supervised implicit language representation is proposed in this study. In comparison with the x-vector representation, the proposed representation provides a relative improvement of $63.9\%$ and achieved a JER of $21.8$ using the E2E framework.

eess.AS↗