Ina Kodrasi

Yacouba Kaloga, Shashi Kumar, Shakeel A. Sheikh et al.

End-to-end ASR systems typically use fixed-depth acoustic encoders at inference, making it difficult to trade additional test-time computation for improved recognition without training a larger model. A natural approach is to reuse a shared Transformer block recurrently, but we find that naive looping does not fully exploit additional recurrent compute. We introduce LARM, a depth-conditioned looped Transformer that turns recurrent encoder depth into a controllable test-time compute axis. LARM combines sparse CTC checkpoints, supervision-clock embeddings, FiLM depth conditioning, and delayed soft-posterior feedback. These components structure the loop into recognition checkpoints separated by latent refinement phases and allow shared weights to specialize across recurrent steps. On LibriSpeech, LARM improves WER as the number of inference loops increases and achieves performance competitive with deeper unshared-parameter baselines. Our results show that test-time compute scaling can extend beyond autoregressive language-model reasoning to continuous non-autoregressive speech recognition.

Shashi Kumar, Yacouba Kaloga, Petr Motlicek et al.

Large language models solve complex problems by generating lengthy chains of explicit reasoning tokens. While effective, this makes reasoning expensive, length-sensitive, and constrained to (discrete) natural language. While latent reasoning offers a continuous alternative, determining useful structures for intermediate latent states is an open challenge. In this paper, we formulate latent reasoning as a geometric path-approximation problem within the model's pretrained token-embedding space. We introduce Geometric Latent Reasoning (GLR), which uses a lightweight transition head to predict iterative direction updates in embedding space. Using textual chain-of-thought traces as anchors, GLR learns to approximate discrete reasoning trajectories while permitting continuous deviations from exact token embeddings. Evaluations on mathematical reasoning benchmarks using Qwen3 models reveal an emergent phenomenon: geometric latent reasoning induces substantially shorter generations without an explicit length objective. By replacing early explicit reasoning with continuous latent steps, models often reach correct answers using substantially fewer total generation steps. These findings suggest that continuous trajectories act as compact intermediate reasoning states, exposing a new tradeoff between latent computation budget, output length, and accuracy.

Shakeel A. Sheikh, Yacouba Kaloga, Md Sahidullah et al.

Despite the promising performance of state of the art approaches for Parkinsons Disease (PD) detection, these approaches often analyze individual speech segments in isolation, which can lead to suboptimal results. Dysarthric cues that characterize speech impairments from PD patients are expected to be related across segments from different speakers. Isolated segment analysis fails to exploit these inter segment relationships. Additionally, not all speech segments from PD patients exhibit clear dysarthric symptoms, introducing label noise that can negatively affect the performance and generalizability of current approaches. To address these challenges, we propose a novel PD detection framework utilizing Graph Convolutional Networks (GCNs). By representing speech segments as nodes and capturing the similarity between segments through edges, our GCN model facilitates the aggregation of dysarthric cues across the graph, effectively exploiting segment relationships and mitigating the impact of label noise. Experimental results demonstrate theadvantages of the proposed GCN model for PD detection and provide insights into its underlying mechanisms

Yacouba Kaloga, Shakeel A. Sheikh, Ina Kodrasi

Recently proposed automatic pathological speech detection approaches rely on spectrogram input representations or wav2vec2 embeddings. These representations may contain pathology irrelevant uncorrelated information, such as changing phonetic content or variations in speaking style across time, which can adversely affect classification performance. To address this issue, we propose to use Multiview Canonical Correlation Analysis (MCCA) on these input representations prior to automatic pathological speech detection. Our results demonstrate that unlike other dimensionality reduction techniques, the use of MCCA leads to a considerable improvement in pathological speech detection performance by eliminating uncorrelated information present in the input representations. Employing MCCA with traditional classifiers yields a comparable or higher performance than using sophisticated architectures, while preserving the representation structure and providing interpretability.

Shashi Kumar, Yacouba Kaloga, John Mitros et al.

Low-rank adaptation (LoRA) is a widely used method for parameter-efficient finetuning. However, existing LoRA variants lack mechanisms to explicitly disambiguate task-relevant information within the learned low-rank subspace, potentially limiting downstream performance. We propose Factorized Variational Autoencoder LoRA (FVAE-LoRA), which leverages a VAE to learn two distinct latent spaces. Our novel Evidence Lower Bound formulation explicitly promotes factorization between the latent spaces, dedicating one latent space to task-salient features and the other to residual information. Extensive experiments on text, audio, and image tasks demonstrate that FVAE-LoRA consistently outperforms standard LoRA. Moreover, spurious correlation evaluations confirm that FVAE-LoRA better isolates task-relevant signals, leading to improved robustness under distribution shifts. Our code is publicly available at: https://github.com/idiap/FVAE-LoRA

Yacouba Kaloga, Shashi Kumar, Petr Motlicek et al.

Accurate sequence-to-sequence (seq2seq) alignment is critical for applications like medical speech analysis and language learning tools relying on automatic speech recognition (ASR). State-of-the-art end-to-end (E2E) ASR systems, such as the Connectionist Temporal Classification (CTC) and transducer-based models, suffer from peaky behavior and alignment inaccuracies. In this paper, we propose a novel differentiable alignment framework based on one-dimensional optimal transport, enabling the model to learn a single alignment and perform ASR in an E2E manner. We introduce a pseudo-metric, called Sequence Optimal Transport Distance (SOTD), over the sequence space and discuss its theoretical properties. Based on the SOTD, we propose Optimal Temporal Transport Classification (OTTC) loss for ASR and contrast its behavior with CTC. Experimental results on the TIMIT, AMI, and LibriSpeech datasets show that our method considerably improves alignment performance compared to CTC and the more recently proposed Consistency-Regularized CTC, though with a trade-off in ASR performance. We believe this work opens new avenues for seq2seq alignment research, providing a solid foundation for further exploration and development within the community.

Shakeel A. Sheikh, Ina Kodrasi

Automatic pathological speech detection approaches yield promising results in identifying various pathologies. These approaches are typically designed and evaluated for phonetically-controlled speech scenarios, where speakers are prompted to articulate identical phonetic content. While gathering controlled speech recordings can be laborious, spontaneous speech can be conveniently acquired as potential patients navigate their daily routines. Further, spontaneous speech can be valuable in detecting subtle and abstract cues of pathological speech. Nonetheless, the efficacy of automatic pathological speech detection for spontaneous speech remains unexplored. This paper analyzes the influence of speech mode on pathological speech detection approaches, examining two distinct categories of approaches, i.e., classical machine learning and deep learning. Results indicate that classical approaches may struggle to capture pathology-discriminant cues in spontaneous speech. In contrast, deep learning approaches demonstrate superior performance, managing to extract additional cues that were previously inaccessible in non-spontaneous speech

Ina Kodrasi

Deep learning-based techniques for automatic dysarthric speech detection have recently attracted interest in the research community. State-of-the-art techniques typically learn neurotypical and dysarthric discriminative representations by processing time-frequency input representations such as the magnitude spectrum of the short-time Fourier transform (STFT). Although these techniques are expected to leverage perceptual dysarthric cues, representations such as the magnitude spectrum of the STFT do not necessarily convey perceptual aspects of complex sounds. Inspired by the temporal processing mechanisms of the human auditory system, in this paper we factor signals into the product of a slowly varying envelope and a rapidly varying fine structure. Separately exploiting the different perceptual cues present in the envelope (i.e., phonetic information, stress, and voicing) and fine structure (i.e., pitch, vowel quality, and breathiness), two discriminative representations are learned through a convolutional neural network and used for automatic dysarthric speech detection. Experimental results show that processing both the envelope and fine structure representations yields a considerably better dysarthric speech detection performance than processing only the envelope, fine structure, or magnitude spectrum of the STFT representation.

Parvaneh Janbakhshi, Ina Kodrasi

Recently proposed automatic pathological speech classification techniques use unsupervised auto-encoders to obtain a high-level abstract representation of speech. Since these representations are learned based on reconstructing the input, there is no guarantee that they are robust to pathology-unrelated cues such as speaker identity information. Further, these representations are not necessarily discriminative for pathology detection. In this paper, we exploit supervised auto-encoders to extract robust and discriminative speech representations for Parkinson's disease classification. To reduce the influence of speaker variabilities unrelated to pathology, we propose to obtain speaker identity-invariant representations by adversarial training of an auto-encoder and a speaker identification task. To obtain a discriminative representation, we propose to jointly train an auto-encoder and a pathological speech classifier. Experimental results on a Spanish database show that the proposed supervised representation learning methods yield more robust and discriminative representations for automatically classifying Parkinson's disease speech, outperforming the baseline unsupervised representation learning system.