scholarly journals Consistent Transcription and Translation of Speech

2020 ◽  
Vol 8 ◽  
pp. 695-709
Author(s):  
Matthias Sperber ◽  
Hendra Setiawan ◽  
Christian Gollan ◽  
Udhyakumar Nallasamy ◽  
Matthias Paulik
Keyword(s):  
Strong Consistency ◽  
High Accuracy ◽  
Inference Procedure ◽  
Speech Translation ◽  
Practical Applications ◽  
Trade Offs ◽  
End To End ◽  
Direct Models ◽  
Translation Accuracy ◽  
Conventional Paradigm

The conventional paradigm in speech translation starts with a speech recognition step to generate transcripts, followed by a translation step with the automatic transcripts as input. To address various shortcomings of this paradigm, recent work explores end-to-end trainable direct models that translate without transcribing. However, transcripts can be an indispensable output in practical applications, which often display transcripts alongside the translations to users. We make this common requirement explicit and explore the task of jointly transcribing and translating speech. Although high accuracy of transcript and translation are crucial, even highly accurate systems can suffer from inconsistencies between both outputs that degrade the user experience. We introduce a methodology to evaluate consistency and compare several modeling approaches, including the traditional cascaded approach and end-to-end models. We find that direct models are poorly suited to the joint transcription/translation task, but that end-to-end models that feature a coupled inference procedure are able to achieve strong consistency. We further introduce simple techniques for directly optimizing for consistency, and analyze the resulting trade-offs between consistency, transcription accuracy, and translation accuracy. 1

scholarly journals End-to-End Sequential Indoor Localization Using Smartphone Inertial Sensors and WiFi

2021 ◽  
Author(s):  
Mukhamet Nurpeiissov ◽  
Askat Kuzdeuov ◽  
Aslan Assylkhanov, ◽  
Yerbolat Khassanov ◽  
Hüseyin Atakan Varol
Keyword(s):  
Indoor Localization ◽  
Inertial Sensors ◽  
Sampling Rate ◽  
Reference Points ◽  
Measurement Unit ◽  
Localization System ◽  
Practical Applications ◽  
External Data ◽  
End To End ◽  
Commercial Off The Shelf

This paper addresses sequential indoor localization using WiFi and Inertial Measurement Unit (IMU) modules commonly found in commercial off-the-shelf smartphones. Specifically, we developed an end-to-end neural network-based localization system integrating WiFi received signal strength indicator (RSSI) and IMU data without external data fusion models. The developed system leverages the advantages of WiFi and IMU modules to locate finer-level sequential positions of a user at 150 Hz sampling rate. Additionally, to demonstrate the efficacy of the proposed approach, we created the IMUWiFine dataset comprising IMU and WiFi RSSI readings sequentially collected at fine-level reference points. The dataset contains 120 trajectories covering an aggregate distance of over 14 kilometers. We conducted extensive experiments using deep learning models and achieved a mean error distance of 1.1 meters on an unseen evaluation set, which makes our approach suitable for many practical applications requiring meter-level accuracy. To enable experiment and result reproducibility, we made the developed localization system and IMUWiFine dataset publicly available in our GitHub repository.<br>

scholarly journals A decision analysis framework for stakeholder involvement and learning in groundwater management

2013 ◽  
Vol 10 (7) ◽  
pp. 8747-8780 ◽  
Author(s):  
T. P. Karjalainen ◽  
P. M. Rossi ◽  
P. Ala-aho ◽  
R. Eskelinen ◽  
K. Reinikainen ◽  
...  
Keyword(s):  
Decision Analysis ◽  
Groundwater Management ◽  
Stakeholder Participation ◽  
Scientific Information ◽  
Stakeholder Involvement ◽  
Resource Planning ◽  
Assessment Process ◽  
Analysis Framework ◽  
Practical Applications ◽  
Trade Offs

Abstract. Multi-criteria decision analysis (MCDA) methods are increasingly used to facilitate both rigorous analysis and stakeholder involvement in natural and water resource planning. Decision making in that context is often complex and multi-faceted with numerous trade-offs between social, environmental and economic impacts. However, practical applications of decision-support methods are often too technically oriented and hard to use, understand or interpret for all participants. The learning of participants in these processes is seldom examined, even though successful deliberation depends on learning. This paper analyzes the potential of an interactive MCDA framework, the decision analysis interview (DAI) approach, for facilitating stakeholder involvement and learning in groundwater management. It evaluates the results of an MCDA process in assessing land-use management alternatives in a Finnish esker aquifer area where conflicting land uses affect the groundwater body and dependent ecosystems. In the assessment process, emphasis was placed on the interactive role of the MCDA tool in facilitating stakeholder participation and learning. The results confirmed that the structured decision analysis framework can foster learning and collaboration in a process where disputes and diverse interests are represented. Computer-aided interviews helped the participants to see how their preferences affected the desirability and ranking of alternatives. During the process, the participants' knowledge and preferences evolved as they assess their initial knowledge with the help of fresh scientific information. The decision analysis process led to the opening of a dialogue, showing the overall picture of the problem context, and the critical issues for the further process.

scholarly journals 923: Avoiding trade-offs: Clinical experience for noninvasive prenatal screen with low no-call rate and high accuracy

2019 ◽  
Vol 220 (1) ◽  
pp. S596-S597
Author(s):  
Susan E. Hancock ◽  
Rotem Ben-Shachar ◽  
Christa Adusei ◽  
Carrie Haverty ◽  
Dale Muzzey
Keyword(s):  
Clinical Experience ◽  
High Accuracy ◽  
Call Rate ◽  
Trade Offs ◽  
Prenatal Screen

scholarly journals LiDAR Integrated High-Capacity IR OWC System with Abilities of Localization and Link Alignment

2021 ◽  
Author(s):  
Zhi Li
Keyword(s):  
Wireless Communication ◽  
Continuous Wave ◽  
Beam Steering ◽  
High Capacity ◽  
High Accuracy ◽  
Signal Beam ◽  
Optical Wireless Communication ◽  
Optical Wireless ◽  
Lidar System ◽  
Practical Applications

By using narrow infrared (IR) beams, optical wireless communication (OWC) system can realize ultra-high capacity and high-privacy transmission. However, due to the point-to-point connection approach, a high-accuracy localization system and beam-steering antenna (BSA) are required to steer the signal beam to user terminals. In addition, to achieve link alignment in the receiver, the BSA needs to be within the limited receiver field of view (FoV). This problem greatly limits the practical application of high-capacity IR optical wireless communication and has not been well solved yet. In this paper, we proposed an indoor beam-steering IR OWC system with high-accuracy and calibration-free localization ability by employing a coaxial frequency modulated continuous wave (FMCW) light detection and ranging (LiDAR) system. In the meantime, benefit from the mm-level ranging accuracy of the LiDAR system, a feasible approach to deal with the link alignment issue for practical applications is firstly demonstrated. With the assistance of the LiDAR system, we experimentally achieved the localization of user terminals with a 0.038-degree localization accuracy, link alignment at the receiver and an error-free on-off keying (OOK) downlink transmission of 20 Gb/s in free space at 3-m distance is demonstrated.

scholarly journals The NiuTrans End-to-End Speech Translation System for IWSLT 2021 Offline Task

2021 ◽  
Author(s):  
Chen Xu ◽  
Xiaoqian Liu ◽  
Xiaowen Liu ◽  
Tiger Wang ◽  
Canan Huang ◽  
...  
Keyword(s):  
Translation System ◽  
Speech Translation ◽  
End To End
Sign in / Sign up

Export Citation Format

Share Document