RCRomit Roy Choudhury

Romit Roy Choudhury

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Romit Roy Choudhury has a total of 14 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Preliminary Technical Validation of LittleBeats™: A Multimodal Sensing Platform to Capture Cardiac Physiology, Motion, and Vocalizations

    by Bashima Islam, Nancy L. McElwain, Jialu Li, Maria I. Davila, Yannan Hu, Kexin Hu, Jordan M. Bodway, Ashutosh Dhekne, Romit Roy Choudhury, Mark Hasegawa-Johnson

    Across five studies, we present the preliminary technical validation of an infant-wearable platform, LittleBeats™, that integrates electrocardiogram (ECG), inertial measurement unit (IMU), and audio sensors. Each sensor modality is validated against data from gold-standard equipment using established algorithms and laboratory tasks. Interbeat interval (IBI) data obtained from the LittleBeats™ ECG sensor indicate acceptable mean absolute percent error rates for both adults (Study 1, N = 16) and infants (Study 2, N = 5) across low- and high-challenge sessions and expected patterns of change in respiratory sinus arrythmia (RSA). For automated activity recognition (upright vs. walk vs. glide vs. squat) using accelerometer data from the LittleBeats™ IMU (Study 3, N = 12 adults), performance was good to excellent, with smartphone (industry standard) data outperforming LittleBeats™ by less than 4 percentage points. Speech emotion recognition (Study 4, N = 8 adults) applied to LittleBeats™ versus smartphone audio data indicated a comparable performance, with no significant difference in error rates. On an automatic speech recognition task (Study 5, N = 12 adults), the best performing algorithm yielded relatively low word error rates, although LittleBeats™ (4.16%) versus smartphone (2.73%) error rates were somewhat higher. Together, these validation studies indicate that LittleBeats™ sensors yield a data quality that is largely comparable to those obtained from gold-standard devices and established protocols used in prior research.


  • Open Access

    Preliminary Technical Validation of LittleBeats™: A Multimodal Sensing Platform to Capture Cardiac Physiology, Motion, and Vocalizations

    by Bashima Islam, Nancy L McElwain, Jialu Li, Maria Davila, Yannan Hu, Kexin Hu, Jordan M Bodway, Ashutosh M Dhekne, Romit Roy Choudhury, Mark Hasegawa-Johnson

    Background: The use of wearable devices has burgeoned over the past decade, including wearables for infants and young children. Typically, such devices assess a single modality, and few have undergone scientific validation. To address this gap, we developed an infant wearable platform, LittleBeats™, that integrates electrocardiogram (ECG), motion, and audio sensors on a single printed circuit board to permit daylong remote assessments of infants and caregivers in home environments. Objective: Complementing our prior reports focused on LittleBeats™ data collected among infants and children in the home context, our main objective here is to present a technical validation of each sensor modality against established laboratory protocols and gold-standard equipment. Materials and Methods: We conducted five studies, including assessments of (a) interbeat interval (IBI) data obtained from the LittleBeats™ ECG sensor versus the gold-standard BIOPAC among adults (Study 1, N=16) and infants (Study 2, N=5), (b) performance of automated activity recognition (upright vs. walk vs. glide vs. squat ) among adults using accelerometer data obtained from LittleBeats™ versus Google Pixel 1 smartphone (Study 3, N=12), and (c) performance of speech emotion recognition (SER; Study 4, N=8) and automatic speech recognition (ASR; Study 5, N =12) algorithms among adult samples using audio data from LittleBeats™ versus smartphone. Results: Results for IBI data obtained from the LittleBeats™ ECG sensor indicate acceptable mean absolute percent error (MAPE) rates for both the adults (MAPE = 5.29% to 5.97%; Study 1) and infants (MAPE =0.96% to 1.66%; Study 2) across low- and high-challenge sessions, as well as expected patterns of change in respiratory sinus arrythmia (RSA) across sessions. For activity recognition (Study 3), LittleBeats™ data showed good to excellent performance (Accuracy = 89%, F1-score=88%; Cohen’s kappa =.79), although the smartphone outperformed LittleBeats™ by less than 4%. Finally, performance on the SER applied to LittleBeats™ versus smartphone audio data (Study 4) indicated comparable performance, with a matched-pairs test indicating no significant difference in error rates (p=.26). On the ASR task (Study 5), the best performing algorithm yielded relatively low word error rates, although LittleBeats™ (4.16%) versus smartphone (2.73%) error rates are somewhat higher. Discussion: Taken together, results from these controlled laboratory studies conducted predominantly with adults indicate that the LittleBeats™ sensors yield data quality that is largely comparable to those obtained from gold-standard devices and protocols that have been used extensively in prior research. The advantages of the LittleBeats™ platform is the (a) integration of multiple sensors into one platform that was (b) designed specifically for use with infants and young children. Leveraging data from the multiple modalities and fine-tuning postprocessing steps to further increase data quality, combined with assessing data quality among infants and young children in their natural environments, will be key next steps in this research program.

  • Open Access

    Learning to Separate Voices by Spatial Regions

    by Zhongweiyang Xu, Romit Roy Choudhury

    We consider the problem of audio voice separation for binaural applications, such as earphones and hearing aids. While today's neural networks perform remarkably well (separating $4+$ sources with 2 microphones) they assume a known or fixed maximum number of sources, K. Moreover, today's models are trained in a supervised manner, using training data synthesized from generic sources, environments, and human head shapes. This paper intends to relax both these constraints at the expense of a slight alteration in the problem definition. We observe that, when a received mixture contains too many sources, it is still helpful to separate them by region, i.e., isolating signal mixtures from each conical sector around the user's head. This requires learning the fine-grained spatial properties of each region, including the signal distortions imposed by a person's head. We propose a two-stage self-supervised framework in which overheard voices from earphones are pre-processed to extract relatively clean personalized signals, which are then used to train a region-wise separation model. Results show promising performance, underscoring the importance of personalization over a generic supervised approach. (audio samples available at our project website: https://uiuc-earable-computing.github.io/binaural/. We believe this result could help real-world applications in selective hearing, noise cancellation, and audio augmented reality.

  • Pervasive Video and Audio

    by Fahim Kawsar, Romit Roy Choudhury, Ganesh Ananthanarayanan

    The articles in this special section focus on pervasive video and audio. For a long time, cameras and microphones have been studied to give machines the ability to see and hear the world around them. The remarkable progress in artificial intelligence (AI) in recent years has pushed this perception ability to a new high. AI-enabled cameras and microphones are now able to replicate human-like reasoning and cognition, be it detecting cancers better than a human doctor, identifying people’s sexual orientation, or understanding human speech. Naturally, these interminable possibilities across a wide range of verticals, ranging from public safety, infrastructure efficiency, entertainment, education, and health, have led to the deployment of billions of cameras and microphones around us—in our homes, workplaces, and cities.

  • If WiFi APs Could Move: A Measurement Study

    by Ashutosh Dhekne, Mahanth Gowda, Romit Roy Choudhury, Srihari Nelakuditi

    This paper explores the possibility of injecting mobility into wireless network infrastructure. We envision WiFi APs on wheels that move to optimize user performance. Movements need not be all around the floor, neither do they have to operate on batteries. As a first step, WiFi APs at home could remain tethered to power and Ethernet outlets while moving in small areas (perhaps under the couch). If such systems prove successful, perhaps future buildings could offer explicit support for network infrastructure mobility. This paper begins with a higher level discussion of robotic wireless networks-the opportunities and the hurdles-and then pivots by developing a smaller slice of the vision through a system called iMob. With iMob, a WiFi AP is mounted on a Roomba robot and made to periodically move within a 2x2 sqft region. The core questions pertain to finding the best location to move to, such that the SNRs from its clients are strong, and the interferences from other APs are weak. Our measurements show that the richness of wireless multipath offers significant opportunities-even within a 2x2 sqft region, locations exist that are 1:7x better than the average location in terms of throughput. When multiple APs in a neighborhood coordinate, the gains can be even higher. In sum, although infrastructure mobility has been discussed in the context of Google Balloons, ad hoc networks, and delay tolerant networks, we believe that the possibility of moving our personal devices in homes and offices is relatively unexplored, and could open doors to new kinds of innovation.