arXiv subjects

Peter Sullivan

Publications and source records attributed to Peter Sullivan.

11 recordsLinked to original sources

arXiv2026

Arab Voices: Mapping Standard and Dialectal Arabic Speech Technology

Dialectal Arabic (DA) speech data vary widely in domain coverage, dialect labeling practices, and recording conditions, complicating cross-dataset comparison and model evaluation. To characterize this landscape, we conduct a computational analysis of linguistic ``dialectness'' alongside objective proxies of audio quality on the training splits of widely used DA corpora. We find substantial heterogeneity both in acoustic conditions and in the strength and consistency of dialectal signals across datasets, underscoring the need for standardized characterization beyond coarse labels. To reduce fragmentation and support reproducible evaluation, we introduce Arab Voices, a standardized framework for DA ASR. Arab Voices provides unified access to 31 datasets spanning 14 dialects, with harmonized metadata and evaluation utilities. We further benchmark a range of recent ASR systems, establishing strong baselines for modern DA ASR.

cs.CL↗

arXiv2025

Quantum NLP models on Natural Language Inference

Quantum natural language processing (QNLP) offers a novel approach to semantic modeling by embedding compositional structure directly into quantum circuits. This paper investigates the application of QNLP models to the task of Natural Language Inference (NLI), comparing quantum, hybrid, and classical transformer-based models under a constrained few-shot setting. Using the lambeq library and the DisCoCat framework, we construct parameterized quantum circuits for sentence pairs and train them for both semantic relatedness and inference classification. To assess efficiency, we introduce a novel information-theoretic metric, Information Gain per Parameter (IGPP), which quantifies learning dynamics independent of model size. Our results demonstrate that quantum models achieve performance comparable to classical baselines while operating with dramatically fewer parameters. The Quantum-based models outperform randomly initialized transformers in inference and achieve lower test error on relatedness tasks. Moreover, quantum models exhibit significantly higher per-parameter learning efficiency (up to five orders of magnitude more than classical counterparts), highlighting the promise of QNLP in low-resource, structure-sensitive settings. To address circuit-level isolation and promote parameter sharing, we also propose a novel cluster-based architecture that improves generalization by tying gate parameters to learned word clusters rather than individual tokens.

cs.CL↗

arXiv2025

NADI 2025: The First Multidialectal Arabic Speech Processing Shared Task

We present the findings of the sixth Nuanced Arabic Dialect Identification (NADI 2025) Shared Task, which focused on Arabic speech dialect processing across three subtasks: spoken dialect identification (Subtask 1), speech recognition (Subtask 2), and diacritic restoration for spoken dialects (Subtask 3). A total of 44 teams registered, and during the testing phase, 100 valid submissions were received from eight unique teams. The distribution was as follows: 34 submissions for Subtask 1 "five teams{\ae}, 47 submissions for Subtask 2 "six teams", and 19 submissions for Subtask 3 "two teams". The best-performing systems achieved 79.8% accuracy on Subtask 1, 35.68/12.20 WER/CER (overall average) on Subtask 2, and 55/13 WER/CER on Subtask 3. These results highlight the ongoing challenges of Arabic dialect speech processing, particularly in dialect identification, recognition, and diacritic restoration. We also summarize the methods adopted by participating teams and briefly outline directions for future editions of NADI.

cs.CL↗

arXiv2025

On Barriers to Archival Audio Processing

In this study, we leverage a unique UNESCO collection of mid-20th century radio recordings to probe the robustness of modern off-the-shelf language identification (LID) and speaker recognition (SR) methods, especially with respect to the impact of multilingual speakers and cross-age recordings. Our findings suggest that LID systems, such as Whisper, are increasingly adept at handling second-language and accented speech. However, speaker embeddings remain a fragile component of speech processing pipelines that is prone to biases related to the channel, age, and language. Issues which will need to be overcome should archives aim to employ SR methods for speaker indexing.

cs.SD↗

arXiv2024

AI Transparency in Academic Search Systems: An Initial Exploration

As AI-enhanced academic search systems become increasingly popular among researchers, investigating their AI transparency is crucial to ensure trust in the search outcomes, as well as the reliability and integrity of scholarly work. This study employs a qualitative content analysis approach to examine the websites of a sample of 10 AI-enhanced academic search systems identified through university library guides. The assessed level of transparency varies across these systems: five provide detailed information about their mechanisms, three offer partial information, and two provide little to no information. These findings indicate that the academic community is recommending and using tools with opaque functionalities, raising concerns about research integrity, including issues of reproducibility and researcher responsibility.

cs.CY↗

arXiv2023

VoxArabica: A Robust Dialect-Aware Arabic Speech Recognition System

Arabic is a complex language with many varieties and dialects spoken by over 450 millions all around the world. Due to the linguistic diversity and variations, it is challenging to build a robust and generalized ASR system for Arabic. In this work, we address this gap by developing and demoing a system, dubbed VoxArabica, for dialect identification (DID) as well as automatic speech recognition (ASR) of Arabic. We train a wide range of models such as HuBERT (DID), Whisper, and XLS-R (ASR) in a supervised setting for Arabic DID and ASR tasks. Our DID models are trained to identify 17 different dialects in addition to MSA. We finetune our ASR models on MSA, Egyptian, Moroccan, and mixed data. Additionally, for the remaining dialects in ASR, we provide the option to choose various models such as Whisper and MMS in a zero-shot setting. We integrate these models into a single web interface with diverse features such as audio recording, file upload, model selection, and the option to raise flags for incorrect outputs. Overall, we believe VoxArabica will be useful for a wide range of audiences concerned with Arabic research. Our system is currently running at https://cdce-206-12-100-168.ngrok.io/.

cs.CL↗

arXiv2023

On the Robustness of Arabic Speech Dialect Identification

Arabic dialect identification (ADI) tools are an important part of the large-scale data collection pipelines necessary for training speech recognition models. As these pipelines require application of ADI tools to potentially out-of-domain data, we aim to investigate how vulnerable the tools may be to this domain shift. With self-supervised learning (SSL) models as a starting point, we evaluate transfer learning and direct classification from SSL features. We undertake our evaluation under rich conditions, with a goal to develop ADI systems from pretrained models and ultimately evaluate performance on newly collected data. In order to understand what factors contribute to model decisions, we carry out a careful human study of a subset of our data. Our analysis confirms that domain shift is a major challenge for ADI models. We also find that while self-training does alleviate this challenges, it may be insufficient for realistic conditions.

eess.AS↗

arXiv2022

Improving Automatic Speech Recognition for Non-Native English with Transfer Learning and Language Model Decoding

ASR systems designed for native English (L1) usually underperform on non-native English (L2). To address this performance gap, \textbf{(i)} we extend our previous work to investigate fine-tuning of a pre-trained wav2vec 2.0 model \cite{baevski2020wav2vec,xu2021self} under a rich set of L1 and L2 training conditions. We further \textbf{(ii)} incorporate language model decoding in the ASR system, along with the fine-tuning method. Quantifying gains acquired from each of these two approaches separately and an error analysis allows us to identify different sources of improvement within our models. We find that while the large self-trained wav2vec 2.0 may be internalizing sufficient decoding knowledge for clean L1 speech \cite{xu2021self}, this does not hold for L2 speech and accounts for the utility of employing language model decoding on L2 data.

cs.CL↗

arXiv2021

Speech Technology for Everyone: Automatic Speech Recognition for Non-Native English with Transfer Learning

To address the performance gap of English ASR models on L2 English speakers, we evaluate fine-tuning of pretrained wav2vec 2.0 models (Baevski et al., 2020; Xu et al., 2021) on L2-ARCTIC, a non-native English speech corpus (Zhao et al., 2018) under different training settings. We compare \textbf{(a)} models trained with a combination of diverse accents to ones trained with only specific accents and \textbf{(b)} results from different single-accent models. Our experiments demonstrate the promise of developing ASR models for non-native English speakers, even with small amounts of L2 training data and even without a language model. Our models also excel in the zero-shot setting where we train on multiple L2 datasets and test on a blind L2 test set.

eess.AS↗

arXiv2014

The Transiting Exoplanet Survey Satellite

The Transiting Exoplanet Survey Satellite (TESS) will search for planets transiting bright and nearby stars. TESS has been selected by NASA for launch in 2017 as an Astrophysics Explorer mission. The spacecraft will be placed into a highly elliptical 13.7-day orbit around the Earth. During its two-year mission, TESS will employ four wide-field optical CCD cameras to monitor at least 200,000 main-sequence dwarf stars with I = 4-13 for temporary drops in brightness caused by planetary transits. Each star will be observed for an interval ranging from one month to one year, depending mainly on the star's ecliptic latitude. The longest observing intervals will be for stars near the ecliptic poles, which are the optimal locations for follow-up observations with the James Webb Space Telescope. Brightness measurements of preselected target stars will be recorded every 2 min, and full frame images will be recorded every 30 min. TESS stars will be 10-100 times brighter than those surveyed by the pioneering Kepler mission. This will make TESS planets easier to characterize with follow-up observations. TESS is expected to find more than a thousand planets smaller than Neptune, including dozens that are comparable in size to the Earth. Public data releases will occur every four months, inviting immediate community-wide efforts to study the new planets. The TESS legacy will be a catalog of the nearest and brightest stars hosting transiting planets, which will endure as highly favorable targets for detailed investigations.

astro-ph.EP↗

arXiv2013

Spitzer Observations of GJ3470b: a Very Low-density Neptune-size Planet Orbiting a Metal-rich M dwarf

We present Spitzer/IRAC 4.5-micron transit photometry of GJ3470b, a Neptune-size planet orbiting a M1.5 dwarf star with a 3.3-day period recently discovered in the course of the HARPS M-dwarf survey. We refine the stellar parameters by employing purely empirical mass-luminosity and surface brightness relations constrained by our updated value for the mean stellar density, and additional information from new near-infrared spectroscopic observations. We derive a stellar mass of M_star = 0.539+0.047-0.043 M_sun and a radius of R_star = 0.568+0.037-0.031 R_sun. We determine the host star of GJ3470b to be metal-rich, with a metallicity of [Fe/H] = +0.20 +/- 0.10 and an effective temperature of Teff = 3600 +/- 100 K. The revised stellar parameters yield a planetary radius R_pl = 4.83+0.22-0.21 R_Earth that is 13 percent larger than the value previously reported in the literature. We find a planetary mass M_pl = 13.9+1.5-1.4 M_Earth that translates to a very low planetary density, rho_pl = 0.72+0.13-0.12 gcm-3, which is 33% smaller than the original value. With a mean density half of that of GJ436b, GJ3470b is an example of a very low-density low-mass planet, similar to Kepler-11d, Kepler-11e, and Kepler-18c but orbiting a much brighter nearby star that is more conducive to follow-up studies.

astro-ph.EP↗