The effect of language models on phonetic decoding for spoken term detection

Spoken term detection (STD) popularly involves performing word or sub-word level speech recognition and indexing the result. This work challenges the assumption that improved speech recognition accuracy implies better indexing for STD. Using an index derived from phone lattices, this paper examines the effect of language model selection on the relationship between phone recognition accuracy and STD accuracy. Results suggest that language models usually improve phone recognition accuracy but their inclusion does not always translate to improved STD accuracy. The findings suggest that using phone recognition accuracy to measure the quality of an STD index can be problematic, and highlight the need for an alternative that is more closely aligned with the goals of the specific detection task.

Spoken term detection using fast phonetic decoding

While spoken term detection (STD) systems based on word indices provide good accuracy, there are several practical applications where it is infeasible or too costly to employ an LVCSR engine. An STD system is presented, which is designed to incorporate a fast phonetic decoding front-end and be robust to decoding errors whilst still allowing for rapid search speeds. This goal is achieved through mono-phone open-loop decoding coupled with fast hierarchical phone lattice search. Results demonstrate that an STD system that is designed with the constraint of a fast and simple phonetic decoding front-end requires a compromise to be made between search speed and search accuracy.

The development of the graphics-decoding proficiency instrument

The Graphics-Decoding Proficiency (G-DP) instrument was developed as a screening test for the purpose of measuring students’ (aged 8-11 years) capacity to solve graphics-based mathematics tasks. These tasks include number lines, column graphs, maps and pie charts. The instrument was developed within a theoretical framework which highlights the various types of information graphics commonly presented to students in large-scale national and international assessments. The instrument provides researchers, classroom teachers and test designers with an assessment tool which measures students’ graphics decoding proficiency across and within five broad categories of information graphics. The instrument has implications for a number of stakeholders in an era where graphics have become an increasingly important way of representing information.

IODP Expedition 339 in the Gulf of Cadiz and off West Iberia : decoding the environmental significance of the Mediterranean outflow water and its global influence

IODP Expedition 339 drilled five sites in the Gulf of Cadiz and two off the west Iberian margin (November 2011 to January 2012), and recovered 5.5 km of sediment cores with an average recovery of 86.4%. The Gulf of Cadiz was targeted for drilling as a key location for the investigation of Mediterranean outflow water (MOW) through the Gibraltar Gateway and its influence on global circulation and climate. It is also a prime area for understanding the effects of tectonic activity on evolution of the Gibraltar Gateway and on margin sedimentation. We penetrated into the Miocene at two different sites and established a strong signal of MOW in the sedimentary record of the Gulf of Cadiz, following the opening of the Gibraltar Gateway. Preliminary results show the initiation of contourite deposition at 4.2–4.5 Ma, although subsequent research will establish whether this dates the onset of MOW. The Pliocene succession, penetrated at four sites, shows low bottom current activity linked with a weak MOW. Significant widespread unconformities, present in all sites but with hiatuses of variable duration, are interpreted as a signal of intensified MOW, coupled with flow confinement. The Quaternary succession shows a much more pronounced phase of contourite drift development, with two periods of MOW intensification separated by a widespread unconformity. Following this, the final phase of drift evolution established the contourite depositional system (CDS) architecture we see today. There is a significant climate control on this evolution of MOW and bottom-current activity. However, from the closure of the Atlantic–Mediterranean gateways in Spain and Morocco just over 6 Ma and the opening of the Gibraltar Gateway at 5.3 Ma, there has been an even stronger tectonic control on margin development, downslope sediment transport and contourite drift evolution. The Gulf of Cadiz is the world's premier contourite laboratory and thus presents an ideal testing ground for the contourite paradigm. Further study of these contourites will allow us to resolve outstanding issues related to depositional processes, drift budgets, and recognition of fossil contourites in the ancient record on shore. The expedition also verified an enormous quantity and extensive distribution of contourite sands that are clean and well sorted. These represent a relatively untapped and important exploration target for potential oil and gas reservoirs.

Brain decoding based on functional magnetic resonance imaging using machine learning : a comparative study

Brain decoding of functional Magnetic Resonance Imaging data is a pattern analysis task that links brain activity patterns to the experimental conditions. Classifiers predict the neural states from the spatial and temporal pattern of brain activity extracted from multiple voxels in the functional images in a certain period of time. The prediction results offer insight into the nature of neural representations and cognitive mechanisms and the classification accuracy determines our confidence in understanding the relationship between brain activity and stimuli. In this paper, we compared the efficacy of three machine learning algorithms: neural network, support vector machines, and conditional random field to decode the visual stimuli or neural cognitive states from functional Magnetic Resonance data. Leave-one-out cross validation was performed to quantify the generalization accuracy of each algorithm on unseen data. The results indicated support vector machine and conditional random field have comparable performance and the potential of the latter is worthy of further investigation.

Water bottles from China: Decoding visual grammar

On the ALEA Study Tour to China, Beryl Exley and her roomie Kathryn O’Sullivan pondered over their first night dilemma whilst staying at a hotel in Beijing. They read the room service guide (in English) which advised against drinking the tap water and confirming the supply of one bottle of complementary water per guest per day. The room service guide listed ‘special’ bottled water was the equivalent of $AUS7 per bottle. However the dilemma was this: sitting on the shelf above the fridge were three different kinds of water-like bottles. Each had a different label, written mainly in Chinese characters. Not wanting to mistake the bottles, Beryl and Kathryn set about decoding the text of the three bottles in question.

An Input Triggered Polymorphic ASIC for H.264 Decoding

This paper reports the design of an input-triggered polymorphic ASIC for H.264 baseline decoder. Hardware polymorphism is achieved by selectively reusing hardware resources at system and module level. Complete design is done using ESL design tools following a methodology that maintains consistency in testing and verification throughout the design flow. The proposed design can support frame sizes from QCIF to 1080p.

Low ML-Decoding Complexity, Large Coding Gain, Full-Rate, Full-Diversity STBCs for 2 x 2 and 4 x 2 MIMO Systems

This paper deals with low maximum-likelihood (ML)-decoding complexity, full-rate and full-diversity space-time block codes (STBCs), which also offer large coding gain, for the 2 transmit antenna, 2 receive antenna (2 x 2) and the 4 transmit antenna, 2 receive antenna (4 x 2) MIMO systems. Presently, the best known STBC for the 2 2 system is the Golden code and that for the 4 x 2 system is the DjABBA code. Following the approach by Biglieri, Hong, and Viterbo, a new STBC is presented in this paper for the 2 x 2 system. This code matches the Golden code in performance and ML-decoding complexity for square QAM constellations while it has lower ML-decoding complexity with the same performance for non-rectangular QAM constellations. This code is also shown to be information-lossless and diversity-multiplexing gain (DMG) tradeoff optimal. This design procedure is then extended to the 4 x 2 system and a code, which outperforms the DjABBA code for QAM constellations with lower ML-decoding complexity, is presented. So far, the Golden code has been reported to have an ML-decoding complexity of the order of for square QAM of size. In this paper, a scheme that reduces its ML-decoding complexity to M-2 root M is presented.

Algebraic distributed space-time codes with low ML decoding complexity

"Extended Clifford algebras" are introduced as a means to obtain low ML decoding complexity space-time block codes. Using left regular matrix representations of two specific classes of extended Clifford algebras, two systematic algebraic constructions of full diversity Distributed Space-Time Codes (DSTCs) are provided for any power of two number of relays. The left regular matrix representation has been shown to naturally result in space-time codes meeting the additional constraints required for DSTCs. The DSTCs so constructed have the salient feature of reduced Maximum Likelihood (ML) decoding complexity. In particular, the ML decoding of these codes can be performed by applying the lattice decoder algorithm on a lattice of four times lesser dimension than what is required in general. Moreover these codes have a uniform distribution of power among the relays and in time, thus leading to a low Peak to Average Power Ratio at the relays.

A non-orthogonal cooperative multiple access (NCMA) protocol and low ML decoding complexity codes

A half-duplex constrained non-orthogonal cooperative multiple access (NCMA) protocol suitable for transmission of information from N users to a single destination in a wireless fading channel is proposed. Transmission in this protocol comprises of a broadcast phase and a cooperation phase. In the broadcast phase, each user takes turn broadcasting its data to all other users and the destination in an orthogonal fashion in time. In the cooperation phase, each user transmits a linear function of what it received from all other users as well as its own data. In contrast to the orthogonal extension of cooperative relay protocols to the cooperative multiple access channels wherein at any point of time, only one user is considered as a source and all the other users behave as relays and do not transmit their own data, the NCMA protocol relaxes the orthogonality built into the protocols and hence allows for a more spectrally efficient usage of resources. Code design criteria for achieving full diversity of N in the NCMA protocol is derived using pair wise error probability (PEP) analysis and it is shown that this can be achieved with a minimum total time duration of 2N - 1 channel uses. Explicit construction of full diversity codes is then provided for arbitrary number of users. Since the Maximum Likelihood decoding complexity grows exponentially with the number of users, the notion of g-group decodable codes is introduced for our setup and a set of necesary and sufficient conditions is also obtained.

Joint decoding of multiple speech patterns for robust speech recognition

We are addressing a new problem of improving automatic speech recognition performance, given multiple utterances of patterns from the same class. We have formulated the problem of jointly decoding K multiple patterns given a single Hidden Markov Model. It is shown that such a solution is possible by aligning the K patterns using the proposed Multi Pattern Dynamic Time Warping algorithm followed by the Constrained Multi Pattern Viterbi Algorithm The new formulation is tested in the context of speaker independent isolated word recognition for both clean and noisy patterns. When 10 percent of speech is affected by a burst noise at -5 dB Signal to Noise Ratio (local), it is shown that joint decoding using only two noisy patterns reduces the noisy speech recognition error rate to about 51 percent, when compared to the single pattern decoding using the Viterbi Algorithm. In contrast a simple maximization of individual pattern likelihoods, provides only about 7 percent reduction in error rate.

Noncoherent low-decoding-complexity space-time codes for wireless relay networks

The differential encoding/decoding setup introduced by Kiran et at, Oggier et al and Jing et al for wireless relay networks that use codebooks consisting of unitary matrices is extended to allow codebooks consisting of scaled unitary matrices. For such codebooks to be used in the Jing-Hassibi protocol for cooperative diversity, the conditions that need to be satisfied by the relay matrices and the codebook are identified. A class of previously known rate one, full diversity, four-group encodable and four-group decodable Differential Space-Time Codes (DSTCs) is proposed for use as Distributed DSTCs (DDSTCs) in the proposed set up. To the best of our knowledge, this is the first known low decoding complexity DDSTC scheme for cooperative wireless networks.

Signal set design for full-diversity low-decoding-complexity differential scaled-unitary STBCs

The problem of designing high rate, full diversity noncoherent space-time block codes (STBCs) with low encoding and decoding complexity is addressed. First, the notion of g-group encodable and g-group decodable linear STBCs is introduced. Then for a known class of rate-1 linear designs, an explicit construction of fully-diverse signal sets that lead to four-group encodable and four-group decodable differential scaled unitary STBCs for any power of two number of antennas is provided. Previous works on differential STBCs either sacrifice decoding complexity for higher rate or sacrifice rate for lower decoding complexity.