arXiv subjects

Raja Kushalnagar

Publications and source records attributed to Raja Kushalnagar.

At least 19 recordsLinked to original sources

arXiv2026

Are Caption Metrics Broken? Latency, Deaf and Hard of Hearing User Ratings, and Bias across Technologies

Live captions on TV often contain errors and timing issues, making it hard for deaf and hard-of-hearing (DHH) viewers to follow dialog. It is essential that caption quality metrics reflect the lived DHH TV viewing experience. To this end, we describe a U.S.-based large-scale online survey with 216 validated participants, who provided 302 responses containing a cumulative 4,832 data points. Participants viewed videos drawn from a pool of 70 clips recorded from live TV, and were asked to rate the caption quality and subjective understanding of the content across four conditions: TV captions as originally recorded with up to 7-12 seconds delay, TV captions synchronized with audio, Automatic Speech Recognition (ASR)-generated captions synchronized with audio, and ASR captions with an average two-second delay. All captions were evaluated against the Word Error Rate (WER), Automated Caption Evaluation (ACE2) and Number, Edition and Recognition (NER) metrics. Results show that TV and ASR captions were rated similarly. For TV captions, all three metrics were moderately-to-highly correlated with viewer ratings, but far less so for ASR captions, making them far from technology-neutral. Additionally, caption latencies significantly impact the viewer experience, especially typical 7-12-second TV delays. We discuss the implications for the adoption of caption quality metrics.

cs.HC↗

arXiv2026

Bigger than the EAR BOX: A Theory-Grounded Review of XR Accessibility Research for Deaf and Hard of Hearing Communities

Extended Reality (XR) technologies have received growing attention in accessibility research involving Deaf and Hard of Hearing (DHH) communities. Yet less attention has been given to the assumptions shaping this work. We present a theory-grounded review of XR research involving DHH users. Drawing on Disability Studies, Deaf Studies, and DeafSpace, we develop a theoretical framework with four analytical dimensions: orientation toward access, distribution of responsibility, conceptualization of DHH communities, and spatial and perceptual assumptions. We apply this framework to 53 XR studies involving DHH users, identified through a search and screening of ACM publications from 2015 to 2025. Our analysis shows that XR accessibility is frequently framed as supporting communication within hearing-default environments, while overlooking the diversity of DHH communities. We identify directions for redistributing accessibility labor and reconfiguring space, and offer our framework as a tool for future XR research involving DHH communities.

cs.HC↗

arXiv2026

DiscoSign: Discourse-Aware Text to Sign Language Gloss Translation

Sign language processing systems have traditionally operated at the sentence level, ignoring critical discourse phenomena fundamental to sign language comprehension. We introduce DiscoSign, a computational approach for discourse-aware text to sign language gloss translation grounded in linguistic research. We address three key phenomena within our modular Large Language Model (LLM)-based translation framework: (i) spatial coreference resolution, where entities maintain consistent spatial locations throughout discourse; (ii) Question-Answer Clauses (QACs), pseudocleft structures serving specific discourse functions; and (iii) concept-gloss consistency, ensuring stable mappings between English concepts and American Sign Language (ASL) signs. Traditional translation metrics fail to capture discourse-level quality, so we introduce a suite of novel evaluation metrics designed to assess each dimension of discourse coherence addressed by our framework. Experiments on sentence-level and discourse-level datasets show that our approach for discourse-aware processing significantly improves spatial consistency and entity tracking relative to sentence-only translation, while maintaining competitive single-sentence gloss translation quality. Our work establishes the first systematic framework for discourse-level text to sign language gloss translation with corresponding evaluation methodology.

cs.CL↗

arXiv2026

Bootstrapping Sign Language Annotations with Sign Language Models

AI-driven sign language interpretation is limited by a lack of high-quality annotated data. New datasets including ASL STEM Wiki and FLEURS-ASL contain professional interpreters and 100s of hours of data but remain only partially annotated and thus underutilized, in part due to the prohibitive costs of annotating at this scale. In this work, we develop a pseudo-annotation pipeline that takes signed video and English as input and outputs a ranked set of likely annotations, including time intervals, for glosses, fingerspelled words, and sign classifiers. Our pipeline uses sparse predictions from our fingerspelling recognizer and isolated sign recognizer (ISR), along with a K-Shot LLM approach, to estimate these annotations. In service of this pipeline, we establish simple yet effective baseline fingerspelling and ISR models, achieving state-of-the-art on FSBoard (6.7% CER) and on ASL Citizen datasets (74% top-1 accuracy). To validate and provide a gold-standard benchmark, a professional interpreter annotated nearly 500 videos from ASL STEM Wiki with sequence-level gloss labels containing glosses, classifiers, and fingerspelling signs. These human annotations and over 300 hours of pseudo-annotations are being released in supplemental material.

cs.CV↗

arXiv2026

Deaf and Hard of Hearing Access to Intelligent Personal Assistants: Comparison of Voice-Based Options with an LLM-Powered Touch Interface

We investigate intelligent personal assistants (IPAs) accessibility for deaf and hard of hearing (DHH) people who can use their voice in everyday communication. The inability of IPAs to understand diverse accents including deaf speech renders them largely inaccessible to non-signing and speaking DHH individuals. Using an Echo Show, we compare the usability of natural language input via spoken English; with Alexa's automatic speech recognition and a Wizard-of-Oz setting with a trained facilitator re-speaking commands against that of a large language model (LLM)-assisted touch interface in a mixed-methods study. The touch method was navigated through an LLM-powered "task prompter," which integrated the user's history and smart environment to suggest contextually-appropriate commands. Quantitative results showed no significant differences across both spoken English conditions vs LLM-assisted touch. Qualitative results showed variability in opinions on the usability of each method. Ultimately, it will be necessary to have robust deaf-accented speech recognized natively by IPAs.

cs.HC↗

arXiv2025

EmoSign: A Multimodal Dataset for Understanding Emotions in American Sign Language

Unlike spoken languages where the use of prosodic features to convey emotion is well studied, indicators of emotion in sign language remain poorly understood, creating communication barriers in critical settings. Sign languages present unique challenges as facial expressions and hand movements simultaneously serve both grammatical and emotional functions. To address this gap, we introduce EmoSign, the first sign video dataset containing sentiment and emotion labels for 200 American Sign Language (ASL) videos. We also collect open-ended descriptions of emotion cues. Annotations were done by 3 Deaf ASL signers with professional interpretation experience. Alongside the annotations, we include baseline models for sentiment and emotion classification. This dataset not only addresses a critical gap in existing sign language research but also establishes a new benchmark for understanding model capabilities in multimodal emotion recognition for sign languages. The dataset is made available at https://huggingface.co/datasets/catfang/emosign.

cs.CV↗

arXiv2025

Towards AI-driven Sign Language Generation with Non-manual Markers

Sign languages are essential for the Deaf and Hard-of-Hearing (DHH) community. Sign language generation systems have the potential to support communication by translating from written languages, such as English, into signed videos. However, current systems often fail to meet user needs due to poor translation of grammatical structures, the absence of facial cues and body language, and insufficient visual and motion fidelity. We address these challenges by building on recent advances in LLMs and video generation models to translate English sentences into natural-looking AI ASL signers. The text component of our model extracts information for manual and non-manual components of ASL, which are used to synthesize skeletal pose sequences and corresponding video frames. Our findings from a user study with 30 DHH participants and thorough technical evaluations demonstrate significant progress and identify critical areas necessary to meet user needs.

cs.HC↗

arXiv2024

"We do use it, but not how hearing people think": How the Deaf and Hard of Hearing Community Uses Large Language Model Tools

Generative AI tools, particularly those utilizing large language models (LLMs), are increasingly used in everyday contexts. While these tools enhance productivity and accessibility, little is known about how Deaf and Hard of Hearing (DHH) individuals engage with them or the challenges they face when using them. This paper presents a mixed-method study exploring how the DHH community uses Text AI tools like ChatGPT to reduce communication barriers and enhance information access. We surveyed 80 DHH participants and conducted interviews with 11 participants. Our findings reveal important benefits, such as eased communication and bridging Deaf and hearing cultures, alongside challenges like lack of American Sign Language (ASL) support and Deaf cultural understanding. We highlight unique usage patterns, propose inclusive design recommendations, and outline future research directions to improve Text AI accessibility for the DHH community.

cs.HC↗

arXiv2024

Customizing Generated Signs and Voices of AI Avatars: Deaf-Centric Mixed-Reality Design for Deaf-Hearing Communication

This study investigates innovative interaction designs for communication and collaborative learning between learners of mixed hearing and signing abilities, leveraging advancements in mixed reality technologies like Apple Vision Pro and generative AI for animated avatars. Adopting a participatory design approach, we engaged 15 d/Deaf and hard of hearing (DHH) students to brainstorm ideas for an AI avatar with interpreting ability (sign language to English, voice to English) that would facilitate their face-to-face communication with hearing peers. Participants envisioned the AI avatars to address some issues with human interpreters, such as lack of availability, and provide affordable options to expensive personalized interpreting service. Our findings indicate a range of preferences for integrating the AI avatars with actual human figures of both DHH and hearing communication partners. The participants highlighted the importance of having control over customizing the AI avatar, such as AI-generated signs, voices, facial expressions, and their synchronization for enhanced emotional display in communication. Based on our findings, we propose a suite of design recommendations that balance respecting sign language norms with adherence to hearing social norms. Our study offers insights on improving the authenticity of generative AI in scenarios involving specific, and sometimes unfamiliar, social norms.

cs.HC↗

arXiv2024

"Real Learner Data Matters" Exploring the Design of LLM-Powered Question Generation for Deaf and Hard of Hearing Learners

Deaf and Hard of Hearing (DHH) learners face unique challenges in learning environments, often due to a lack of tailored educational materials that address their specific needs. This study explores the potential of Large Language Models (LLMs) to generate personalized quiz questions to enhance DHH students' video-based learning experiences. We developed a prototype leveraging LLMs to generate questions with emphasis on two unique strategies: Visual Questions, which identify video segments where visual information might be misrepresented, and Emotion Questions, which highlight moments where previous DHH learners experienced learning difficulty manifested in emotional responses. Through user studies with DHH undergraduates, we evaluated the effectiveness of these LLM-generated questions in supporting the learning experience. Our findings indicate that while LLMs offer significant potential for personalized learning, challenges remain in the interaction accessibility for the diverse DHH community. The study highlights the importance of considering language diversity and culture in LLM-based educational technology design.

cs.HC↗

arXiv2024

Assessment of Sign Language-Based versus Touch-Based Input for Deaf Users Interacting with Intelligent Personal Assistants

With the recent advancements in intelligent personal assistants (IPAs), their popularity is rapidly increasing when it comes to utilizing Automatic Speech Recognition within households. In this study, we used a Wizard-of-Oz methodology to evaluate and compare the usability of American Sign Language (ASL), Tap to Alexa, and smart home apps among 23 deaf participants within a limited-domain smart home environment. Results indicate a slight usability preference for ASL. Linguistic analysis of the participants' signing reveals a diverse range of expressions and vocabulary as they interacted with IPAs in the context of a restricted-domain application. On average, deaf participants exhibited a vocabulary of 47 +/- 17 signs with an additional 10 +/- 7 fingerspelled words, for a total of 246 different signs and 93 different fingerspelled words across all participants. We discuss the implications for the design of limited-vocabulary applications as a stepping-stone toward general-purpose ASL recognition in the future.

cs.HC↗

arXiv2024

Sign Language-Based versus Touch-Based Input for Deaf Users with Interactive Personal Assistants in Simulated Kitchen Environments

In this study, we assess the usability of interactive personal assistants (IPAs), such as Amazon Alexa, in a simulated kitchen smart home environment, with deaf and hard of hearing users. Participants engage in activities in a way that causes their hands to get dirty. With these dirty hands, they are tasked with two different input methods for IPAs: American Sign Language (ASL) in a Wizard-of-Oz design, and smart home apps with a touchscreen. Usability ratings show that participants significantly preferred ASL over touch-based apps with dirty hands, although not to a larger extent than in comparable previous work with clean hands. Participants also expressed significant enthusiasm for ASL-based IPA interaction in Netpromoter scores and in questions about their overall preferences. Preliminary observations further suggest that having dirty hands may affect the way people sign, which may pose challenges for building IPAs that natively support sign language input.

cs.HC↗

arXiv2024

How Users Experience Closed Captions on Live Television: Quality Metrics Remain a Challenge

This paper presents a mixed methods study on how deaf, hard of hearing and hearing viewers perceive live TV caption quality with captioned video stimuli designed to mirror TV captioning experiences. To assess caption quality, we used four commonly-used quality metrics focusing on accuracy: word error rate, weighted word error rate, automated caption evaluation (ACE), and its successor ACE2. We calculated the correlation between the four quality metrics and viewer ratings for subjective quality and found that the correlation was weak, revealing that other factors besides accuracy affect user ratings. Additionally, even high-quality captions are perceived to have problems, despite controlling for confounding factors. Qualitative analysis of viewer comments revealed three major factors affecting their experience: Errors within captions, difficulty in following captions, and caption appearance. The findings raise questions as to how objective caption quality metrics can be reconciled with the user experience across a diverse spectrum of viewers.

cs.HC↗

arXiv2022

Live Captions in Virtual Reality (VR)

Few VR applications and games implement captioning of speech and audio cues, which either inhibits or prevents access of their application by deaf or hard of hearing (DHH) users, new language learners, and other caption users. Additionally, little to no guidelines exist on how to implement live captioning on VR headsets and how it may differ from traditional television captioning. To help fill the void of information behind user preferences of different VR captioning styles, we conducted a study with eight DHH participants to test three caption movement behaviors (headlocked, lag, and appear) while watching live-captioned, single-speaker presentations in VR. Participants answered a series of Likert scale and open-ended questions about their experience. Participant preferences were split, but the majority of participants reported feeling comfortable with using live captions in VR and enjoyed the experience. When participants ranked the caption behaviors, there was almost an equal divide between the three types tested. IPQ results indicated each behavior had similar immersion ratings, however participants found headlocked and lag captions more user-friendly than appear captions. We suggest that participants may vary in caption preference depending on how they use captions, and that providing opportunities for caption customization is best.

cs.HC↗

arXiv2021

Social, Environmental, and Technical: Factors at Play in the Current Use and Future Design of Small-Group Captioning

Real-time captioning is a critical accessibility tool for many d/Deaf and hard of hearing (DHH) people. While the vast majority of captioning work has focused on formal settings and technical innovations, in contrast, we investigate captioning for informal, interactive small-group conversations, which have a high degree of spontaneity and foster dynamic social interactions. This paper reports on semi-structured interviews and design probe activities we conducted with 15 DHH participants to understand their use of existing real-time captioning services and future design preferences for both in-person and remote small-group communication. We found that our participants' experiences of captioned small-group conversations are shaped by social, environmental, and technical considerations (e.g., interlocutors' pre-established relationships, the type of captioning displays available, and how far captions lag behind speech). When considering future captioning tools, participants were interested in greater feedback on non-speech elements of conversation (e.g., speaker identity, speech rate, volume) both for their personal use and to guide hearing interlocutors toward more accessible communication. We contribute a qualitative account of DHH people's real-time captioning experiences during small-group conversation and future design considerations to better support the groups being captioned, both in person and online.

cs.HC↗

arXiv2019

RTTD-ID: Tracked Captions with Multiple Speakers for Deaf Students

Students who are deaf and hard of hearing cannot hear in class and do not have full access to spoken information. They can use accommodations such as captions that display speech as text. However, compared with their hearing peers, the caption accommodations do not provide equal access, because they are focused on reading captions on their tablet and cannot see who is talking. This viewing isolation contributes to student frustration and risk of doing poorly or withdrawing from introductory engineering courses with lab components. It also contributes to their lack of inclusion and sense of belonging. We report on the evaluation of a Real-Time Text Display with Speaker-Identification, which displays the location of a speaker in a group (RTTD-ID). RTTD-ID aims to reduce frustration in identifying and following an active speaker when there are multiple speakers, e.g., in a lab. It has three different display schemes to identify the location of the active speaker, which helps deaf students in viewing both the speaker's words and the speaker's expression and actions. We evaluated three RTTD speaker identification methods: 1) traditional: captions stay in one place and viewers search for the speaker, 2) pointer: captions stay in one place, and a pointer to the speaker is displayed, and 3) pop-up: captions "pop-up" next to the speaker. We gathered both quantitative and qualitative information through evaluations with deaf and hard of hearing users. The users preferred the pointer identification method over the traditional and pop-up methods.

cs.HC↗

arXiv2019

Closed ASL Interpreting for Online Videos

Deaf individuals face great challenges in today's society. It can be very difficult to be able to understand different forms of media without a sense of hearing. Many videos and movies found online today are not captioned, and even fewer have a supporting video with an interpreter. Also, even with a supporting interpreter video provided, information is still lost due to the inability to look at both the video and the interpreter simultaneously. To alleviate this issue, we came up with a tool called closed interpreting. Similar to closed captioning, it will be displayed with an online video and can be toggled on and off. However, the closed interpreter is also user-adjustable. Settings, such as interpreter size, transparency, and location, can be adjusted. Our goal with this study is to find out what deaf and hard of hearing viewers like about videos that come with interpreters, and whether the adjustability is beneficial.

cs.HC↗

arXiv2019

Deaf, Hard of Hearing, and Hearing Perspectives on using Automatic Speech Recognition in Conversation

Many personal devices have transitioned from visual-controlled interfaces to speech-controlled interfaces to reduce costs and interactive friction, supported by the rapid growth in capabilities of speech-controlled interfaces, e.g., Amazon Echo or Apple's Siri. A consequence is that people who are deaf or hard of hearing (DHH) may be unable to use these speech-controlled devices. We show that deaf speech has a high error rate compared to hearing speech, in commercial speech-controlled interfaces. Deaf speech had approximately a 78% word error rate (WER) compared to a hearing speech 18% WER. Our findings show that current speech-controlled interfaces are not usable by DHH people. Based on our findings, significant advances in speech recognition software or alternative approaches will be needed for deaf use of speech-controlled interfaces. We show that current speech-controlled interfaces are not usable by DHH people.

cs.HC↗