An Energy-Efficient Speech-Extraction Processor for Robust User Speech Recognition in Mobile Head-Mounted Display Systems

2017 ◽  
Vol 64 (4) ◽  
pp. 457-461 ◽  
Author(s):  
Jinmook Lee ◽  
Seongwook Park ◽  
Injoon Hong ◽  
Hoi-Jun Yoo
Keyword(s):  
Speech Recognition ◽  
Energy Efficient ◽  
Head Mounted Display ◽  
Speech Extraction ◽  
Display Systems

scholarly journals Approximate LSTM Computing for Energy-Efficient Speech Recognition

Electronics ◽  
2020 ◽  
Vol 9 (12) ◽  
pp. 2004
Author(s):  
Junseo Jo ◽  
Jaeha Kung ◽  
Youngjoo Lee
Keyword(s):  
Speech Recognition ◽  
Energy Efficient ◽  
Short Term Memory ◽  
Recognition System ◽  
Similarity Score ◽  
Cmos Process ◽  
Approximate Computing ◽  
Processing Elements ◽  
On Chip ◽  
Computing Method

This paper presents an approximate computing method of long short-term memory (LSTM) operations for energy-efficient end-to-end speech recognition. We newly introduce the concept of similarity score, which can measure how much the inputs of two adjacent LSTM cells are similar to each other. Then, we disable the highly-similar LSTM operations and directly transfer the prior results for reducing the computational costs of speech recognition. The pseudo-LSTM operation is additionally defined for providing the approximate computation with reduced processing resolution, which can further relax the processing overheads without degrading the accuracy. In order to verify the proposed idea, in addition, we design an approximate LSTM accelerator in 65 nm CMOS process. The proposed accelerator newly utilizes a number of approximate processing elements (PEs) to support the proposed skipped-LSTM and pseudo-LSTM operations without degrading the energy efficiency. Moreover, sparsity-aware scheduling is introduced by introducing the small-sized on-chip SRAM buffer. As a result, the proposed work provides an energy-efficient but still accurate speech recognition system, which consumes 2.19 times less energy than the baseline architecture.

Evaluating Effectiveness of Interaction Techniques across Immersive Virtual Environmental Systems

2005 ◽  
Vol 14 (5) ◽  
pp. 511-527 ◽  
Author(s):  
Anthony Steed ◽  
Chris Parker
Keyword(s):  
Field Of View ◽  
Ray Casting ◽  
Interaction Techniques ◽  
Head Mounted Display ◽  
Virtual Hand ◽  
Immersive Systems ◽  
Selection Tasks ◽  
Display Systems ◽  
Simple Selection ◽  
Selection Of

This paper presents the results of a formal experiment to compare different interaction techniques across two types of immersive display: an immersive projection technology (IPT) and a head-mounted display (HMD). Our aim is to investigate the effectiveness of two widely used interaction metaphors, virtual hand and ray casting, on these two display technologies. Our motivation is that design and evaluation of interaction techniques for immersive egocentric display systems has been undertaken almost exclusively on HMDs. We argue that basing interaction for IPTs on techniques developed for other types of immersive systems is a flawed approach, as there are some categorical differences between the experience given by an IPT and an HMD. For example, an IPT user has a much wider field of view than an HMD user. We have chosen two types of interaction tasks to study: simple selection of objects both near to and at some distance from the user, and manipulation of objects involving a change of both position and orientation. As previous studies have found, we find that ray casting is preferable for selection and virtual hand is preferable for manipulation for a HMD. We show that this is also the case for the IPT. More interestingly, while we find performance on selection tasks is much better on the IPT, for manipulation tasks there is little difference between the two display technologies.

scholarly journals Exocentric Distance Judgment and Accuracy of Head-Mounted and Stereoscopic Widescreen Displays in Frontal Planes

2020 ◽  
Vol 10 (4) ◽  
pp. 1427
Author(s):  
Chiuhsiang Joe Lin ◽  
Betsha Tizazu Abreham ◽  
Dino Caesaron ◽  
Bereket Haile Woldegiorgis
Keyword(s):  
Frontal Plane ◽  
Distance Judgment ◽  
Head Mounted Display ◽  
Position Accuracy ◽  
Significant Difference ◽  
Head Mounted Displays ◽  
Position Data ◽  
Exocentric Distance ◽  
3D Applications ◽  
Display Systems

An experiment was done to explore the effects of two virtual display systems on the accuracy of exocentric distance judgment and position. Sixteen participants viewed animated virtual targets using either a head-mounted display (HMD) or a stereoscopic widescreen display (SWD). The virtual targets have been shown, one at a time, at three depth levels and with two corresponding exocentric distances and three target sizes at each target distance and, afterward, via pointing by holding a stick to estimate the exocentric distance and position of each target. The position data were collected using an OptiTrack motion capture system. The results showed that the accuracy of exocentric distance judgment was higher with the head-mounted displays than with the stereoscopic widescreen displays. In addition, higher position accuracy in the X-direction was obtained from the stereoscopic widescreen displays, whereas no significant difference was observed in position accuracy in the Y-direction. However, it is possible that the HMD could give better accuracy in both exocentric distance and position judgments in the frontal plane, if the HMD had been perfectly mounted and flawlessly fit the participant’s eyes. The result also revealed that exocentric distance judgment was significantly higher at the farthest target distances than at the nearest distance. Similarly, the position accuracy significantly increased as exocentric distance decreased. Moreover, engineers may allude to the findings as the evidence from the study suggests that the intermediate target distances might be fitting or ideal distances to design and structure 3D applications.

Sign in / Sign up

Export Citation Format

Share Document