Enhanced Multilingual Speaker Recognition through the Incorporation of High-Level Features, Late Fusion and Discriminative Classification Methods. The development of robust multilingual speaker recognition systems will benefit the community through the elimination of fraud incurred by financial institutions and customers by enabling several person authentication applications such as: voice based signatures and document issuance; credit card verification by voice and secure over-the-phone financi ....Enhanced Multilingual Speaker Recognition through the Incorporation of High-Level Features, Late Fusion and Discriminative Classification Methods. The development of robust multilingual speaker recognition systems will benefit the community through the elimination of fraud incurred by financial institutions and customers by enabling several person authentication applications such as: voice based signatures and document issuance; credit card verification by voice and secure over-the-phone financial transactions. The technology will also assist in the protection of the community and safeguard Australia by enabling the implementation of the following: suspect identification using voice print; national security measures for combating terrorism by using voice to locate and track terrorists; preemptive criminal activity counter-measures; surveillance and secure building access by voice.Read moreRead less
Robust speaker recognition with reduced utterance duration and intersession variability. The development of robust and accurate speaker recognition systems will enable secure person authentication in over-the-phone financial transactions and benefit the community through the elimination of identity fraud incurred by customers and financial institutions. The technology will also assist in safeguarding Australia by enabling the implementation of suspect identification using voice and security meas ....Robust speaker recognition with reduced utterance duration and intersession variability. The development of robust and accurate speaker recognition systems will enable secure person authentication in over-the-phone financial transactions and benefit the community through the elimination of identity fraud incurred by customers and financial institutions. The technology will also assist in safeguarding Australia by enabling the implementation of suspect identification using voice and security measures for combating terrorism by using voice to locate and track terrorists. Our research at QUT Speech Research Lab is at the forefront of development in this field and will provide Australia with a technological advantage in the rapidly evolving global market for speaker recognition technology for person authentication applications.Read moreRead less
Robust Automatic Speaker Diarisation of Audio Documents by Exploiting Prior Sources of Information. Speaker Diarisation, the task of determining who spoke when, is a technology fundamental in deriving intelligent information from audio and multimedia resources. The requirement for efficient and accurate Speaker Diarisation systems, portable across different domains is heightened by the explosive growth of audio and multimedia archives online and throughout the world. This research will provide t ....Robust Automatic Speaker Diarisation of Audio Documents by Exploiting Prior Sources of Information. Speaker Diarisation, the task of determining who spoke when, is a technology fundamental in deriving intelligent information from audio and multimedia resources. The requirement for efficient and accurate Speaker Diarisation systems, portable across different domains is heightened by the explosive growth of audio and multimedia archives online and throughout the world. This research will provide the foundation for a commercial service of automatic Speaker Diarisation to be developed, growing Australia's impact on the information and communications technology (ICT) sector. The outcome of this research will also assist in the tracking of terrorist and unlawful activity by enabling effective intelligence gathering from different audio sources.Read moreRead less
Making demonstrably reliable forensic voice comparison a practical everyday reality in Australia. To assist Australian law-enforcement agencies and courts in the process of the conviction of the guilty and the exoneration of the innocent, this project will develop and test a practical and demonstrably reliable forensic voice comparison system for use with Australian voices. This will allow forensic scientists to produce reliable strength of evidence statements for presentation in court using the ....Making demonstrably reliable forensic voice comparison a practical everyday reality in Australia. To assist Australian law-enforcement agencies and courts in the process of the conviction of the guilty and the exoneration of the innocent, this project will develop and test a practical and demonstrably reliable forensic voice comparison system for use with Australian voices. This will allow forensic scientists to produce reliable strength of evidence statements for presentation in court using the same evaluative framework as used with DNA. In addition, application of the system during criminal investigations may lead to the refocussing of investigations on other suspects, or may help leverage guilty pleas, thus saving substantial time and money.Read moreRead less
Audio Visual Speech Recognition. Even though significant advances have been made in automatic speech recognition using acoustic information, the recognition accuracies are still poor in noisy and hostile environments such as in crowds, traffic, factory floors etc. In many of these applications visual information is or can easily be made available in addition to the audio. The aim of this project is to achieve an order of magnitude improvement in speech recognition accuracies in adverse environme ....Audio Visual Speech Recognition. Even though significant advances have been made in automatic speech recognition using acoustic information, the recognition accuracies are still poor in noisy and hostile environments such as in crowds, traffic, factory floors etc. In many of these applications visual information is or can easily be made available in addition to the audio. The aim of this project is to achieve an order of magnitude improvement in speech recognition accuracies in adverse environments by joint processing and modelling of the acoustic modality with visual information in the form of lip shapes and movements. The outcomes will be useful in human computer interaction in adverse environments as well as in the transcription and mining of multimedia data.
Read moreRead less
Fixed and variable-length segment vocoders for very low bitrate speech coding. Reliable and secure voice communication is an important aspect of military and defence operations. In order to reduce the possibility of interception, low power transmitters are normally used for radio communications, where the bandwidth is often very low. Military voice communication, therefore, requires the coding of speech at very low bitrates. Our research proposal aims to develop speech coders that can operate ....Fixed and variable-length segment vocoders for very low bitrate speech coding. Reliable and secure voice communication is an important aspect of military and defence operations. In order to reduce the possibility of interception, low power transmitters are normally used for radio communications, where the bandwidth is often very low. Military voice communication, therefore, requires the coding of speech at very low bitrates. Our research proposal aims to develop speech coders that can operate at lower bitrates and reproduce speech of high quality and intelligibility. This is highly beneficial to the defence forces of Australia as it will permit the use of high-grade encryption technology to improve the security of transmission.Read moreRead less
Automatic audio segmentation, classification, identification, search and retrieval. The research aims to develop generic tools for automated audio segmentation, classification, identification and search, with lowest possible computational complexity and highest accuracy and speed. The tools will be applicable to audio archive management, search of audio material over WWW and personal archives of music and audio-assisted video analysis. The industry will use the tools for automated broadcast ve ....Automatic audio segmentation, classification, identification, search and retrieval. The research aims to develop generic tools for automated audio segmentation, classification, identification and search, with lowest possible computational complexity and highest accuracy and speed. The tools will be applicable to audio archive management, search of audio material over WWW and personal archives of music and audio-assisted video analysis. The industry will use the tools for automated broadcast verification and identification for copyright surveillance and calculation of royalty payments, aiming to penetrate both Australian and overseas markets. The area of real-time audio scene analysis is in its infancy and the research aims to make significant contributions to this area.Read moreRead less
Detection and Quantification of General Fetal Movements from Accelerometer Measurements using Nonstationary Signal Processing Techniques. There are approximately 1,750 fetal deaths per year in Australian with about one-third occurring late in gestation and without an apparent cause. The development of an automated system capable of long-term monitoring of fetal health will result in accurate diagnoses and prediction of future outcome. This will, in turn, allow early intervention by the clinicia ....Detection and Quantification of General Fetal Movements from Accelerometer Measurements using Nonstationary Signal Processing Techniques. There are approximately 1,750 fetal deaths per year in Australian with about one-third occurring late in gestation and without an apparent cause. The development of an automated system capable of long-term monitoring of fetal health will result in accurate diagnoses and prediction of future outcome. This will, in turn, allow early intervention by the clinician to reduce fetal deaths and enhance the chances of good outcomes with resultant savings in social and financial costs to the community. The development of such equipment would spawn future research into intervention treatments and contribute to Australia's position as a world leader in computerised health monitoring systems.Read moreRead less
Multi-Channel Time-Frequency Analysis for EEG Neonatal Seizure Characterization. This project researches new signal processing methodologies for a multi-channel characterization of seizures for use in diagnosing newborn brain dysfunctions. The outcomes will result in important immediate clinical benefits for sick newborn babies and will fundamentally facilitate research progress in the development of neuroprotectants and anticonvulsants. The success of this project will contribute in minimizing ....Multi-Channel Time-Frequency Analysis for EEG Neonatal Seizure Characterization. This project researches new signal processing methodologies for a multi-channel characterization of seizures for use in diagnosing newborn brain dysfunctions. The outcomes will result in important immediate clinical benefits for sick newborn babies and will fundamentally facilitate research progress in the development of neuroprotectants and anticonvulsants. The success of this project will contribute in minimizing the social financial costs by diagnosing brain disorders in the initial stage of life and preventing further damage. This has the potential to result in a standard diagnostic equipment in neonatal intensive care units and medical research centres.Read moreRead less
Design of Neonatal Seizure Diagnosis Methods Using Time-Frequency Signal Processing Techniques. Seizures occur in approximately 0.5% of all newborns. They are often the only indicator of an early dysfunction in central nervous system (CNS). Their occurrence raises concerns about the underlying cause, its effect on the brain, and the appropriate treatment. Newborn seizures are mostly sub-clinical and only detected through the Electroencephalogram. For an efficient diagnosis, seizure classificatio ....Design of Neonatal Seizure Diagnosis Methods Using Time-Frequency Signal Processing Techniques. Seizures occur in approximately 0.5% of all newborns. They are often the only indicator of an early dysfunction in central nervous system (CNS). Their occurrence raises concerns about the underlying cause, its effect on the brain, and the appropriate treatment. Newborn seizures are mostly sub-clinical and only detected through the Electroencephalogram. For an efficient diagnosis, seizure classification systems were proposed based on visual observations. This project proposes developing a novel approach to automate the classification process using time-frequency (TF) signal processing techniques based on the multi-channel characteristics of the seizure; namely: A) TF signature B) origin, and C) propagation behaviour.Read moreRead less