Sergio Burdisso

Shashi Kumar, Srikanth Madikeri, Juan Zuluaga-Gomez et al.

In traditional conversational intelligence from speech, a cascaded pipeline is used, involving tasks such as voice activity detection, diarization, transcription, and subsequent processing with different NLP models for tasks like semantic endpointing and named entity recognition (NER). Our paper introduces TokenVerse, a single Transducer-based model designed to handle multiple tasks. This is achieved by integrating task-specific tokens into the reference text during ASR model training, streamlining the inference and eliminating the need for separate NLP models. In addition to ASR, we conduct experiments on 3 different tasks: speaker change detection, endpointing, and NER. Our experiments on a public and a private dataset show that the proposed method improves ASR by up to 7.7% in relative WER while outperforming the cascaded pipeline approach in individual task performance. Our code is publicly available: https://github.com/idiap/tokenverse-unifying-speech-nlp

Sergio Burdisso, Séverin Baroudi, Yanis Labrak et al.

We present SDialog, an MIT-licensed open-source Python toolkit that unifies dialog generation, evaluation and mechanistic interpretability into a single end-to-end framework for building and analyzing LLM-based conversational agents. Built around a standardized \texttt{Dialog} representation, SDialog provides: (1) persona-driven multi-agent simulation with composable orchestration for controlled, synthetic dialog generation, (2) comprehensive evaluation combining linguistic metrics, LLM-as-a-judge and functional correctness validators, (3) mechanistic interpretability tools for activation inspection and steering via feature ablation and induction, and (4) audio generation with full acoustic simulation including 3D room modeling and microphone effects. The toolkit integrates with all major LLM backends, enabling mixed-backend experiments under a unified API. By coupling generation, evaluation, and interpretability in a dialog-centric architecture, SDialog enables researchers to build, benchmark and understand conversational systems more systematically.

Sergio Burdisso, Esaú Villatoro-Tello, Srikanth Madikeri et al.

We propose a simple approach for weighting self-connecting edges in a Graph Convolutional Network (GCN) and show its impact on depression detection from transcribed clinical interviews. To this end, we use a GCN for modeling non-consecutive and long-distance semantics to classify the transcriptions into depressed or control subjects. The proposed method aims to mitigate the limiting assumptions of locality and the equal importance of self-connections vs. edges to neighboring nodes in GCNs, while preserving attractive features such as low computational cost, data agnostic, and interpretability capabilities. We perform an exhaustive evaluation in two benchmark datasets. Results show that our approach consistently outperforms the vanilla GCN model as well as previously reported results, achieving an F1=0.84 on both datasets. Finally, a qualitative analysis illustrates the interpretability capabilities of the proposed approach and its alignment with previous findings in psychology.

Sergio Burdisso, Juan Zuluaga-Gomez, Esau Villatoro-Tello et al.

In this paper, we describe our participation in the subtask 1 of CASE-2022, Event Causality Identification with Casual News Corpus. We address the Causal Relation Identification (CRI) task by exploiting a set of simple yet complementary techniques for fine-tuning language models (LMs) on a small number of annotated examples (i.e., a few-shot configuration). We follow a prompt-based prediction approach for fine-tuning LMs in which the CRI task is treated as a masked language modeling problem (MLM). This approach allows LMs natively pre-trained on MLM problems to directly generate textual responses to CRI-specific prompts. We compare the performance of this method against ensemble techniques trained on the entire dataset. Our best-performing submission was fine-tuned with only 256 instances per class, 15.7% of the all available data, and yet obtained the second-best precision (0.82), third-best accuracy (0.82), and an F1-score (0.85) very close to what was reported by the winner team (0.86).

Martin Fajcik, Muskaan Singh, Juan Zuluaga-Gomez et al.

In this paper, we describe our shared task submissions for Subtask 2 in CASE-2022, Event Causality Identification with Casual News Corpus. The challenge focused on the automatic detection of all cause-effect-signal spans present in the sentence from news-media. We detect cause-effect-signal spans in a sentence using T5 -- a pre-trained autoregressive language model. We iteratively identify all cause-effect-signal span triplets, always conditioning the prediction of the next triplet on the previously predicted ones. To predict the triplet itself, we consider different causal relationships such as cause$\rightarrow$effect$\rightarrow$signal. Each triplet component is generated via a language model conditioned on the sentence, the previous parts of the current triplet, and previously predicted triplets. Despite training on an extremely small dataset of 160 samples, our approach achieved competitive performance, being placed second in the competition. Furthermore, we show that assuming either cause$\rightarrow$effect or effect$\rightarrow$cause order achieves similar results.

Iuliia Thorbecke, Esaú Villatoro-Tello, Juan Zuluaga-Gomez et al.

Despite the success of end-to-end automatic speech recognition (ASR) models, challenges persist in recognizing rare, out-of-vocabulary words - including named entities (NE) - and in adapting to new domains using only text data. This work presents a practical approach to address these challenges through an unexplored combination of an NE bias list and a word-level n-gram language model (LM). This solution balances simplicity and effectiveness, improving entities' recognition while maintaining or even enhancing overall ASR performance. We efficiently integrate this enriched biasing method into a transducer-based ASR system, enabling context adaptation with almost no computational overhead. We present our results on three datasets spanning four languages and compare them to state-of-the-art biasing strategies. We demonstrate that the proposed combination of keyword biasing and n-gram LM improves entity recognition by up to 32% relative and reduces overall WER by up to a 12% relative.

Iuliia Thorbecke, Juan Zuluaga-Gomez, Esaú Villatoro-Tello et al.

The training of automatic speech recognition (ASR) with little to no supervised data remains an open question. In this work, we demonstrate that streaming Transformer-Transducer (TT) models can be trained from scratch in consumer and accessible GPUs in their entirety with pseudo-labeled (PL) speech from foundational speech models (FSM). This allows training a robust ASR model just in one stage and does not require large data and computational budget compared to the two-step scenario with pre-training and fine-tuning. We perform a comprehensive ablation on different aspects of PL-based streaming TT models such as the impact of (1) shallow fusion of n-gram LMs, (2) contextual biasing with named entities, (3) chunk-wise decoding for low-latency streaming applications, and (4) TT overall performance as the function of the FSM size. Our results demonstrate that TT can be trained from scratch without supervised data, even with very noisy PLs. We validate the proposed framework on 6 languages from CommonVoice and propose multiple heuristics to filter out hallucinated PLs.

Thibault Bañeras-Roux, Sergio Burdisso, Esaú Villatoro-Tello et al.

Conventional end-to-end automatic speech recognition (ASR) systems rely on paired speech-text data for domain adaptation. Recent LLM-based ASR architectures connect a speech encoder to a large language model via a projection module, enabling adaptation with text-only data. However, this introduces a modality gap, as the LLM is not exposed to the noisy representations produced by the speech projector. We investigate whether small amounts of speech can mitigate this mismatch. We compare three strategies: text-only adaptation, paired speech-text adaptation, and mixed batching (MB), which combines both. Experiments in in-domain and out-of-domain settings show that even limited speech consistently improves performance. Notably, MB using only 10% of the target-domain (less than 4 hours) speech achieves word error rates comparable to, or better than, conventional ASR fine-tuning with the full dataset, indicating that small amounts of speech provide a strong modality-alignment signal.

Hasindri Watawana, Sergio Burdisso, Diego A. Moreno-Galván et al.

Automatic depression detection from doctor-patient conversations has gained momentum thanks to the availability of public corpora and advances in language modeling. However, interpretability remains limited: strong performance is often reported without revealing what drives predictions. We analyze three datasets: ANDROIDS, DAIC-WOZ, E-DAIC and identify a systematic bias from interviewer prompts in semi-structured interviews. Models trained on interviewer turns exploit fixed prompts and positions to distinguish depressed from control subjects, often achieving high classification scores without using participant language. Restricting models to participant utterances distributes decision evidence more broadly and reflects genuine linguistic cues. While semi-structured protocols ensure consistency, including interviewer prompts inflates performance by leveraging script artifacts. Our results highlight a cross-dataset, architecture-agnostic bias and emphasize the need for analyses that localize decision evidence by time and speaker to ensure models learn from participants' language.

Shashi Kumar, Esaú Villatoro-Tello, Sergio Burdisso et al.

Standard LLM-based speech recognition systems typically process utterances in isolation, limiting their ability to leverage conversational context. In this work, we study whether multimodal context from prior turns improves LLM-based ASR and how to represent that context efficiently. We find that, after supervised multi-turn training, conversational context mainly helps with the recognition of contextual entities. However, conditioning on raw context is expensive because the prior-turn audio token sequence grows rapidly with conversation length. To address this, we propose Abstract Compression, which replaces the audio portion of prior turns with a fixed number of learned latent tokens while retaining corresponding transcripts explicitly. On both in-domain and out-of-domain test sets, the compressed model recovers part of the gains of raw-context conditioning with a smaller prior-turn audio footprint. We also provide targeted analyses of the compression setup and its trade-offs.

Sergio Burdisso, Ernesto Reyes-Ramírez, Esaú Villatoro-Tello et al.

Automatic depression detection from conversational data has gained significant interest in recent years. The DAIC-WOZ dataset, interviews conducted by a human-controlled virtual agent, has been widely used for this task. Recent studies have reported enhanced performance when incorporating interviewer's prompts into the model. In this work, we hypothesize that this improvement might be mainly due to a bias present in these prompts, rather than the proposed architectures and methods. Through ablation experiments and qualitative analysis, we discover that models using interviewer's prompts learn to focus on a specific region of the interviews, where questions about past experiences with mental health issues are asked, and use them as discriminative shortcuts to detect depressed participants. In contrast, models using participant responses gather evidence from across the entire interview. Finally, to highlight the magnitude of this bias, we achieve a 0.90 F1 score by intentionally exploiting it, the highest result reported to date on this dataset using only textual information. Our findings underline the need for caution when incorporating interviewers' prompts into models, as they may inadvertently learn to exploit targeted prompts, rather than learning to characterize the language and behavior that are genuinely indicative of the patient's mental health condition.

Shashi Kumar, Iuliia Thorbecke, Sergio Burdisso et al.

Recent research has demonstrated that training a linear connector between speech foundation encoders and large language models (LLMs) enables this architecture to achieve strong ASR capabilities. Despite the impressive results, it remains unclear whether these simple approaches are robust enough across different scenarios and speech conditions, such as domain shifts and speech perturbations. In this paper, we address these questions by conducting various ablation experiments using a recent and widely adopted approach called SLAM-ASR. We present novel empirical findings that offer insights on how to effectively utilize the SLAM-ASR architecture across a wide range of settings. Our main findings indicate that SLAM-ASR exhibits poor performance in cross-domain evaluation settings. Additionally, speech perturbations on in-domain data, such as changes in speech rate or additive noise, can significantly degrade performance. Our findings offer critical insights for fine-tuning and configuring robust LLM-based ASR models, tailored to different data characteristics and computational resources.

Thibault Bañeras-Roux, Shashi Kumar, Driss Khalil et al.

Automatic Speech Recognition (ASR) is traditionally evaluated using Word Error Rate (WER), a metric that is insensitive to meaning. Embedding-based semantic metrics are better correlated with human perception, but decoder-based Large Language Models (LLMs) remain underexplored for this task. This paper evaluates their relevance through three approaches: (1) selecting the best hypothesis between two candidates, (2) computing semantic distance using generative embeddings, and (3) qualitative classification of errors. On the HATS dataset, the best LLMs achieve 92--94\% agreement with human annotators for hypothesis selection, compared to 63\% for WER, also outperforming semantic metrics. Embeddings from decoder-based LLMs show performance comparable to encoder models. Finally, LLMs offer a promising direction for interpretable and semantic ASR evaluation.

Sergio Burdisso, Dairazalia Sánchez-Cortés, Esaú Villatoro-Tello et al.

Evaluating the reliability of news sources is a routine task for journalists and organizations committed to acquiring and disseminating accurate information. Recent research has shown that predicting sources' reliability represents an important first-prior step in addressing additional challenges such as fake news detection and fact-checking. In this paper, we introduce a novel approach for source reliability estimation that leverages reinforcement learning strategies for estimating the reliability degree of news sources. Contrary to previous research, our proposed approach models the problem as the estimation of a reliability degree, and not a reliability label, based on how all the news media sources interact with each other on the Web. We validated the effectiveness of our method on a news media reliability dataset that is an order of magnitude larger than comparable existing datasets. Results show that the estimated reliability degrees strongly correlates with journalists-provided scores (Spearman=0.80) and can effectively predict reliability labels (macro-avg. F$_1$ score=81.05). We release our implementation and dataset, aiming to provide a valuable resource for the NLP community working on information verification.

Sergio Burdisso, Srikanth Madikeri, Petr Motlicek

Efficiently deriving structured workflows from unannotated dialogs remains an underexplored and formidable challenge in computational linguistics. Automating this process could significantly accelerate the manual design of workflows in new domains and enable the grounding of large language models in domain-specific flowcharts, enhancing transparency and controllability. In this paper, we introduce Dialog2Flow (D2F) embeddings, which differ from conventional sentence embeddings by mapping utterances to a latent space where they are grouped according to their communicative and informative functions (i.e., the actions they represent). D2F allows for modeling dialogs as continuous trajectories in a latent space with distinct action-related regions. By clustering D2F embeddings, the latent space is quantized, and dialogs can be converted into sequences of region/action IDs, facilitating the extraction of the underlying workflow. To pre-train D2F, we build a comprehensive dataset by unifying twenty task-oriented dialog datasets with normalized per-turn action annotations. We also introduce a novel soft contrastive loss that leverages the semantic information of these actions to guide the representation learning process, showing superior performance compared to standard supervised contrastive loss. Evaluation against various sentence embeddings, including dialog-specific ones, demonstrates that D2F yields superior qualitative and quantitative results across diverse domains.

Sergio Burdisso, Esaú Villatoro-Tello, Andrés Carofilis et al.

Adapting automatic speech recognition (ASR) systems based on large language models (LLMs) to new domains using text-only data is a significant yet underexplored challenge. Standard fine-tuning of the LLM on target-domain text often disrupts the critical alignment between speech and text modalities learned by the projector, degrading performance. We introduce a novel text-only adaptation method that emulates the audio projection task by treating it as a text denoising task. Our approach thus trains the LLM to recover clean transcripts from noisy inputs. This process effectively adapts the model to a target domain while preserving cross-modal alignment. Our solution is lightweight, requiring no architectural changes or additional parameters. Extensive evaluation on two datasets demonstrates up to 22.1% relative improvement, outperforming recent state-of-the-art text-only adaptation methods.

Yanis Labrak, David Grünert, Séverin Baroudi et al.

Long-context audio reasoning is underserved in both training data and evaluation. Existing benchmarks target short-context tasks, and the open-ended generation tasks most relevant to long-context reasoning pose well-known challenges for automatic evaluation. We propose a synthetic data generation pipeline designed to serve both as a training resource and as a controlled evaluation environment, and instantiate it for first-visit doctor-patient conversations with SOAP note generation as the task. The pipeline has three stages, persona-driven dialogue generation, multi-speaker audio synthesis with overlap/pause modeling, room acoustics, and sound events, and LLM-based reference SOAP note production, built entirely on open-weight models. We release 8,800 synthetic conversations with 1.3k hours of corresponding audio and reference notes. Evaluating current open-weight systems, we find that cascaded approaches still substantially outperform end-to-end models.

Sergio Burdisso, Esaú Villatoro-Tello, Petr Motlicek

The advancement of conversational AI systems relies on the availability of high-quality, flexible, and reproducible synthetic dialogues for training, evaluation, and benchmarking. SDialog is a modular, extensible Python toolkit designed to address the challenges of synthetic dialogue generation and analysis. By leveraging instruction-tuned Large Language Models (LLMs), SDialog provides abstractions for personas, orchestration, and scenario management, enabling the creation of realistic, diverse, and controllable conversational data for research and development. SDialog supports workflows such as multi-agent simulation and scenario-driven generation, and represents a step forward in the standardization of tools and frameworks for synthetic data generation, a crucial advancement for ensuring reproducibility in today's fast-evolving research landscape.

Pradeep Rangappa, Andres Carofilis, Jeena Prakash et al.

Fine-tuning pretrained ASR models for specific domains is challenging for small organizations with limited labeled data and computational resources. Here, we explore different data selection pipelines and propose a robust approach that improves ASR adaptation by filtering pseudo-labels generated using Whisper (encoder-decoder) and Zipformer (transducer) models. Our approach integrates multiple selection strategies -- including word error rate (WER) prediction, named entity recognition (NER), and character error rate (CER) analysis -- to extract high-quality training segments. We evaluate our method on Whisper and Zipformer using a 7500-hour baseline, comparing it to a CER-based approach relying on hypotheses from three ASR systems. Fine-tuning on 7500 hours of pseudo-labeled call center data achieves 12.3% WER, while our filtering reduces the dataset to 100 hours (1.4%) with similar performance; a similar trend is observed on Fisher English.

Dairazalia Sánchez-Cortés, Sergio Burdisso, Esaú Villatoro-Tello et al.

Bias assessment of news sources is paramount for professionals, organizations, and researchers who rely on truthful evidence for information gathering and reporting. While certain bias indicators are discernible from content analysis, descriptors like political bias and fake news pose greater challenges. In this paper, we propose an extension to a recently presented news media reliability estimation method that focuses on modeling outlets and their longitudinal web interactions. Concretely, we assess the classification performance of four reinforcement learning strategies on a large news media hyperlink graph. Our experiments, targeting two challenging bias descriptors, factual reporting and political bias, showed a significant performance improvement at the source media level. Additionally, we validate our methods on the CLEF 2023 CheckThat! Lab challenge, outperforming the reported results in both, F1-score and the official MAE metric. Furthermore, we contribute by releasing the largest annotated dataset of news source media, categorized with factual reporting and political bias labels. Our findings suggest that profiling news media sources based on their hyperlink interactions over time is feasible, offering a bird's-eye view of evolving media landscapes.

Sergio Burdisso, Esaú Villatoro-Tello, Shashi Kumar et al.

LLM-based automatic speech recognition (ASR), a well-established approach, connects speech foundation models to large language models (LLMs) through a speech-to-LLM projector, yielding promising results. A common design choice in these architectures is the use of a fixed, manually defined prompt during both training and inference. This setup not only enables applicability across a range of practical scenarios, but also helps maximize model performance. However, the impact of prompt design remains underexplored. This paper presents a comprehensive analysis of commonly used prompts across diverse datasets, showing that prompt choice significantly affects ASR performance and introduces instability, with no single prompt performing best across all cases. Inspired by the speech-to-LLM projector, we propose a prompt projector module, a simple, model-agnostic extension that learns to project prompt embeddings to more effective regions of the LLM input space, without modifying the underlying LLM-based ASR model. Experiments on four datasets show that the addition of a prompt projector consistently improves performance, reduces variability, and outperforms the best manually selected prompts.

Shashi Kumar, Srikanth Madikeri, Esaú Villatoro-Tello et al.

Token-based multitasking frameworks like TokenVerse require all training utterances to have labels for all tasks, hindering their ability to leverage partially annotated datasets and scale effectively. We propose TokenVerse++, which introduces learnable vectors in the acoustic embedding space of the XLSR-Transducer ASR model for dynamic task activation. This core mechanism enables training with utterances labeled for only a subset of tasks, a key advantage over TokenVerse. We demonstrate this by successfully integrating a dataset with partial labels, specifically for ASR and an additional task, language identification, improving overall performance. TokenVerse++ achieves results on par with or exceeding TokenVerse across multiple tasks, establishing it as a more practical multitask alternative without sacrificing ASR performance.

Andres Carofilis, Pradeep Rangappa, Srikanth Madikeri et al.

Fine-tuning pretrained ASR models for specific domains is challenging when labeled data is scarce. But unlabeled audio and labeled data from related domains are often available. We propose an incremental semi-supervised learning pipeline that first integrates a small in-domain labeled set and an auxiliary dataset from a closely related domain, achieving a relative improvement of 4% over no auxiliary data. Filtering based on multi-model consensus or named entity recognition (NER) is then applied to select and iteratively refine pseudo-labels, showing slower performance saturation compared to random selection. Evaluated on the multi-domain Wow call center and Fisher English corpora, it outperforms single-step fine-tuning. Consensus-based filtering outperforms other methods, providing up to 22.3% relative improvement on Wow and 24.8% on Fisher over single-step fine-tuning with random selection. NER is the second-best filter, providing competitive performance at a lower computational cost.