Dialogue models falter in noisy, multi-speaker environments, often producing irrelevant responses and awkward turn-taking. We present AV-Dialog, the first multimodal dialog framework that uses both audio and visual cues to track the target speaker, predict turn-taking, and generate coherent responses. By combining acoustic tokenization with multi-task, multi-stage training on monadic, synthetic, and real audio-visual dialogue datasets, AV-Dialog achieves robust streaming transcription, semantically grounded turn-boundary detection and accurate responses, resulting in a natural conversational flow. Experiments show that AV-Dialog outperforms audio-only models under interference, reducing transcription errors, improving turn-taking prediction, and enhancing human-rated dialogue quality. These results highlight the power of seeing as well as hearing for speaker-aware interaction, paving the way for {spoken} dialogue agents that perform {robustly} in real-world, noisy environments.

We introduce proactive hearing assistants 1 that automatically identify and separate the wearer's conversation partners, without requiring explicit prompts.Our system operates on egocentric binaural audio and uses the wearer's self-speech as an anchor, leveraging turn-taking behavior and dialogue dynamics to infer conversational partners and suppress others.To enable real-time, on-device operation, we propose a dual-model architecture: a lightweight streaming model runs every 12.5 ms for lowlatency extraction of the conversation partners, while a slower model runs less frequently to capture longer-range conversational dynamics.Results on real-world 2-and 3-speaker conversation test sets, collected with binaural egocentric hardware from 11 participants totaling 6.8 hours, show generalization in identifying and isolating conversational partners in multi-conversation settings.Our work marks a step toward hearing assistants that adapt proactively to conversational dynamics and engagement.

The conventional wisdom has been that designing ultra-compact, battery-constrained wireless hearables with on-device speech AI models is challenging due to the high computational demands of streaming deep learning models. Speech AI models require continuous, real-time audio processing, imposing strict computational and I/O constraints.

In crowded settings, the human brain can focus on speech from a target speaker, given prior knowledge of how they sound. We introduce a novel intelligent hearable system that achieves this capability, enabling target speech hearing to ignore all interfering speech and noise, but the target speaker. A naïve approach is to require a clean speech example to enroll the target speaker. This is however not well aligned with the hearable application domain since obtaining a clean example is challenging in real world scenarios, creating a unique user interface problem. We present the first enrollment interface where the wearer looks at the target speaker for a few seconds to capture a single, short, highly noisy, binaural example of the target speaker. This noisy example is used for enrollment and subsequent speech extraction in the presence of interfering speakers and noise. Our system achieves a signal quality improvement of 7.01 dB using less than 5 seconds of noisy enrollment audio and can process 8 ms of audio chunks in 6.24 ms on an embedded CPU. Our user studies demonstrate generalization to real-world static and mobile speakers in previously unseen indoor and outdoor multipath environments. Finally, our enrollment interface for noisy examples does not cause performance degradation compared to clean examples, while being convenient and user-friendly. Taking a step back, this paper takes an important step towards enhancing the human auditory perception with artificial intelligence.

Extracting the speech of participants in a conversation amidst interfering speakers and noise presents a challenging problem. In this paper, we introduce the novel task of target conversation extraction, where the goal is to extract the audio of a target conversation based on the speaker embedding of one of its participants. To accomplish this, we propose leveraging temporal patterns inherent in human conversations, particularly turn-taking dynamics, which uniquely characterize speakers engaged in conversation and distinguish them from interfering speakers and noise. Using neural networks, we show the feasibility of our approach on English and Mandarin conversation datasets. In the presence of interfering speakers, our results show an 8.19 dB improvement in signal-to-noise ratio for 2-speaker conversations and a 7.92 dB improvement for 2-4-speaker conversations. Code, dataset available at https://github.com/chentuochao/Target-Conversation-Extraction.

We present the first neural network model to achieve real-time and streaming target sound extraction. To accomplish this, we propose Waveformer, an encoder-decoder architecture with a stack of dilated causal convolution layers as the encoder, and a transformer decoder layer as the decoder. This hybrid architecture uses dilated causal convolutions for processing large receptive fields in a computationally efficient manner, while also leveraging the generalization performance of transformer-based architectures. Our evaluations show as much as 2.2–3.3 dB improvement in SI-SNRi compared to the prior models for this task while having a 1.2–4x smaller model size and a 1.5–2x lower runtime. We provide code, dataset, and audio samples: https://waveformer.cs.washington.edu/.

The emergence of water-proof mobile and wearable devices (e.g., Garmin Descent and Apple Watch Ultra) designed for underwater activities like professional scuba diving, opens up opportunities for underwater networking and localization capabilities on these devices. Here, we present the first underwater acoustic positioning system for smart devices. Unlike conventional systems that use floating buoys as anchors at known locations, we design a system where a dive leader can compute the relative positions of all other divers, without any external infrastructure. Our intuition is that in a well-connected network of devices, if we compute the pairwise distances, we can determine the shape of the network topology. By incorporating orientation information about a single diver who is in the visual range of the leader device, we can then estimate the positions of all the remaining divers, even if they are not within sight. We address various practical problems including detecting erroneous distance estimates, addressing rotational and flipping ambiguities as well as designing a distributed timestamp protocol that scales linearly with the number of devices. Our evaluations show that our distributed system running on underwater deployments of 4--5 commodity smart devices can perform pairwise ranging and localization with median errors of 0.5--0.9 m and 0.9--1.6 m. Project page with code: https://underwatergps.cs.washington.edu/

Full-body tracking in virtual reality improves presence, allows interaction via body postures, and facilitates better social expression among users. However, full-body tracking systems today require a complex setup fixed to the environment (e.g., multiple lighthouses/cameras) and a laborious calibration process, which goes against the desire to make VR systems more portable and integrated. We present HybridTrak, which provides accurate, real-time full-body tracking by augmenting inside-out1 upper-body VR tracking systems with a single external off-the-shelf RGB web camera. HybridTrak uses a full-neural solution to convert and transform users’ 2D full-body poses from the webcam to 3D poses leveraging the inside-out upper-body tracking data. We showed HybridTrak is more accurate than RGB or depth-based tracking methods on the MPI-INF-3DHP dataset. We also tested HybridTrak in the popular VRChat app and showed that body postures presented by HybridTrak are more distinguishable and more natural than a solution using an RGBD camera.

In this MobiSys demo, we present the demo of our SIGCOMM 2022 paper on underwater messaging system for existing mobile devices like smartphones and smart watches. Our software-only solution leverages audio sensors, i.e., microphones and speakers, ubiquitous in today's devices to enable acoustic underwater communication between mobile devices. To achieve this, we design a communication system that in real-time adapts to differences in frequency responses across mobile devices, changes in multipath and noise levels at different locations and dynamic channel changes due to mobility. Our demo will allow MobiSys attendees to test our system in realtime in a water tank using several demo smart devices in a waterproof pouch. We will also distribute the Android executable of the system via a QR code to attendees who wish to install and try the system on their own smart devices.

Since its inception, underwater digital acoustic communication has required custom hardware that neither has the economies of scale nor is pervasive. We present the first acoustic system that brings underwater messaging capabilities to existing mobile devices like smartphones and smart watches. Our software-only solution leverages audio sensors, i.e., microphones and speakers, ubiquitous in today's devices to enable acoustic underwater communication between mobile devices. To achieve this, we design a communication system that in real-time adapts to differences in frequency responses across mobile devices, changes in multipath and noise levels at different locations and dynamic channel changes due to mobility. We evaluate our system in six different real-world underwater environments with depths of 2--15 m in the presence of boats, ships and people fishing and kayaking. Our results show that our system can in real-time adapt its frequency band and achieve bit rates of 100 bps to 1.8 kbps and a range of 30 m. By using a lower bit rate of 10--20 bps, we can further increase the range to 100 m. As smartphones and watches are increasingly being used in underwater scenarios, our software-based approach has the potential to make underwater messaging capabilities widely available to anyone with a mobile device.

C-Face (Contour-Face) is an ear-mounted wearable sensing technology that uses two miniature cameras to continuously reconstruct facial expressions by deep learning contours of the face. When facial muscles move, the contours of the face change from the point of view of the ear-mounted cameras. These subtle changes are fed into a deep learning model which continuously outputs 42 facial feature points representing the shapes and positions of the mouth, eyes and eyebrows. To evaluate C-Face, we embedded our technology into headphones and earphones. We conducted a user study with nine participants. In this study, we compared the output of our system to the feature points outputted by a state of the art computer vision library (Dlib) from a font facing camera. We found that the mean error of all 42 feature points was 0.77 mm for earphones and 0.74 mm for headphones. The mean error for 20 major feature points capturing the most active areas of the face was 1.43 mm for earphones and 1.39 mm for headphones. The ability to continuously reconstruct facial expressions introduces new opportunities in a variety of applications. As a demonstration, we implemented and evaluated C-Face for two applications: facial expression detection (outputting emojis) and silent speech recognition. We further discuss the opportunities and challenges of deploying C-Face in real-world applications.