Xianren Zhang, Shreyas Prasad, Di Wang, Qiuhai Zeng, Suhang Wang, Wenbo Yan, Mat Hans. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2026.

Conventional dynamic language switching enables seamless multilingual interactions by running several monolingual ASR systems in parallel and triggering the appropriate downstream components using a standalone language identification (LID) service. Since this solution is neither scalable nor cost- and memory-efficient, especially for on-device applications, we propose end-to-end, streaming, joint ASR-LID architectures based on the recurrent neural network transducer framework. Two key formulations are explored: (1) joint training using a unified output space for ASR and LID vocabularies, and (2) joint training viewed as multi-task optimization. We also evaluate the benefit of using auxiliary language information obtained on-the-fly from an acoustic LID classifier. Experiments with the English-Hindi language pair show that: (a) multi-task architectures perform better overall, and (b) the best joint architecture surpasses monolingual ASR (6.4–9.2% word error rate reduction) and acoustic LID (53.9–56.1% error rate reduction) baselines while reducing the overall memory footprint by up to 46%.

The EncounterEngine platform computes relative user proximity based on Bluetooth scan data. The platform is application-agnostic, works with conventional mobile handsets, is energy-efficient, and is easy to use. We describe the lessons learned in building the EncounterEngine platform and a mobile proximity-based game called "wherepsilas blue?" that runs on it.

We describe and analyze a comprehensive learning model to address issues such as consistency, convergence rate, and sample complexity in the general context of system identification. The learning model is based on unbounded loss functions, and it incorporates a measure of algorithmic deficiency. We define and use a novel formulation of algorithmic solution that is an extension of the empirical risk minimization method in the sense that it uses a generic notion of side information as opposed to the commonly used input/output observation of a system. Sufficient conditions for consistency as well as closed form expressions for exponential convergence rate and sample complexity of the identification algorithm are derived

Named for the combination of the words DJ and Jamming, the DJammer allows its users to manipulate their music using standard DJ techniques as well as interact with others in virtual jam sessions through the exchange and sharing of multiple music streams. In this paper, we describe the evolution of the DJammer including the interactive process and evaluation which led to the latest prototype. This latest prototype allows its users to "touch" their digital music through single-handed manipulation and control. By allowing both creativity and communication with digital music, the DJammer is a new musical instrument which takes the next step in the evolution of portable music players.

We present a new model of interactive audio for entertainment and communication. A new device called the DJammer and its associated technologies are described. The DJammer introduces the idea of provisioning mobile users to interact cooperatively with digital audio streams. Users can augment the audio in real time and communicate the result in several ways resulting in a new form of multimedia communication across diverse devices and multiple networks. This paper describes the technologies incorporated into the DJammer, and discusses the actual implementation of the prototype DJammer. Future enhancements are also described.

This work describes the optimization of a signal processing front-end for a distributed speech recognition system with the goal of reducing power consumption. Two categories of source code optimizations were used, architectural and algorithmic. Architectural optimizations reduce the power consumption for a particular system, in this case, the HP Labs Smartbadge IV prototype portable system. Algorithmic optimizations are more general and involve changes in the algorithmic implementation of the source code to run faster and consume less power. A cycle accurate energy simulation shows a reduction in power usage by 83.5% with these optimizations. The optimized source code runs 34 times faster than the original code, therefore it can run at lower processor clock speeds and voltages for further reductions in power consumption. This technique, known as dynamic voltage scaling, was implemented on the Smartbadge IV hardware for an overall reduction in power usage of 89.2%.

We present a multi-resolution space carving algorithm that reconstructs a 3D model of a visual scene photographed by a calibrated digital camera placed at multiple viewpoints. Our approach employs a level set framework for reconstructing the scene. Unlike most standard space carving approaches, our level set approach produces a smooth reconstruction composed of manifold surfaces. Our method outputs a polygonal model, instead of a collection of voxels. We texture-map the reconstructed geometry using the photographs, and then render the model to produce photo-realistic new views of the scene.

We designed a simple, lossless audio codec, called AudioPaK, which uses only a small number of integer arithmetic operations on both the coder and the decoder side. The main operations of this codec are polynomial prediction and Golomb-Rice coding, and are done on a frame basis. Our coder performs as well, or even better than most lossless audio codecs.

We report the design and implementation of a compliant MPEG audio decoder for Layers I and II with new algorithmic and architectural enhancements. The proposed algorithm uses a scaled 32-point Chen DCT and is implemented using 16-bit integer arithmetic. Measurements and results are given for a complete decoder implemented and optimized for Intel's MMX technology.

A new 16-bit integer MPEG-1 layer II audio decoding algorithm is introduced, this algorithm incorporates a scaled Chen (1977) discrete cosine transform (DCT) instead of a generic DCT in the matrixing block, the use of a scaled DCT reduces the DCT multiply count by 28%. This 16-bit integer decoder is compliant with the Moving Picture Expert Group (MPEG) standard.