1. CASA-Based Robust Speaker Identification(Computational Auditory Scene Analysis)
2. Independent component analysis and MLLRtransformation for speaker identification• Independent Component Analysis (ICA).• Principle Component Analysis (PCA).3. Towards noise –robust speaker recognitionusing probabilistic linear discriminant analysis• Probabilistic linear discriminant analysis• Additive noise
4. Weighted LDA techniques for i-vectorbased speaker verification.• Improving i-vector speaker verification in presence ofhigh inter session variability.• Interview-interview condition.• Telephone-telephone condition.5. An Overview of Speaker Identification:Accuracy and Robustness IssuesTwo methods• Speaker identification• Speaker verification
6. Cross-pollination of normalization techniquesfrom speaker to face authentication usingGaussian mixture models.
7.Front-End Factor Analysis for SpeakerVerification• This paper proposed new way of combining JFA andSVM’s for speaker verification.8. Parallel transformation network feature forspeaker recognition• TN features with SVM modeling-method in order tobecome language independent and overcome the needfor accurate speech recognition.
9. Statistical Pattern Recognition Techniques forSpeaker Verification10. Speaker Identification within Whispered SpeechAudio Streams Whisper is an alternative speech production mode used bysubjects in natural conversation to protect the privacy. Whispered speech is a natural mode of speech information.
11. A comparison of approaches for modeling prosodicfeatures in speaker recognition.• It address the task of text-independent speakerverification.• Prosodic features.12.Fusion Methods for Boosting Performance ofSpeaker Identification Systems1. feature extraction.2.classification tasks.
13. Source-normalized LDA for robust speakerrecognition using i-vectors from multiple speechsources• Improves the robustness of i-vector-based speakerrecognition.• An source-normalized algorithm to improvesrobustness of i-vector-based-speaker recognition.14. A study on Universal Background Model trainingin Speaker Verification• Systematic analyze of speaker verification systemperformance.• Rigorous methods like IFS scheme is used toestimate similarity.
15. Speaker Identification Using InstantaneousFrequencies• Introduction of new set of descriptors that capture theidentity of speaker well.• Provides robustness with respect to changes inrecording channel and speaking style.16. Codebook Design Method for Noise RobustSpeaker Identification based on GeneticAlgorithm• To designing a codebook for noise robust speaker,Genetic algorithm is proposed.
Paradigm of the proposed codebookdesign method.
17. Enhanced speaker recognition based on intra-modal fusion and accent modeling.• Intra-modal fusion.• Accent modeling.18. Discriminant NAP for SVM Speaker Recognition• Nuisance Attribute Projection (NAP) provides aneffective method of removing the unwanted sessionvariability in a Support Vector Machine (SVM) basedspeaker recognition system by removing the principalcomponents of this variability.
19. A Speech-and-Speaker Identification System:Feature Extraction, Description and Classification ofSpeech-Signal Image• A speech-and-speaker (SAS) identification systembased on spoken Arabic digit recognition.
20. In-Set/Out-of-Set Speaker Recognition UnderSparse Enrollment• The problem of in-set speaker recognition isaddressed with the constraints of low enrollment (5 s)and test material (2–8 s) and in-set group sizesranging from 15–45 speakers.• An algorithm is proposed that uses an in-setspeaker’s cohort set to make up for the sparse (e.g., 5s per speaker) enrollment data.21. Analysis of Speech Recognition Techniques foruse in a Non-Speech Sound Recognition System• Analysis the different techniques used for speechrecognition and identifies those that can be used fornon-speech sound recognition
22. Speaker verification for home security system• A reliable speaker verification algorithm is used inhome security.
23. An Efficient Scoring Algorithm for GaussianMixture Model Based Speaker Identification• The use of GMM for speaker identification wasshown to provide superior performanceGraphical illustration of the observation vector recording
24. Speaker Recognition: A Tutorial• Speech processing is a diverse field with manyapplications.
25. Speaker Identification Based on the Use ofRobust Cepstral Features Obtained from Pole-ZeroTransfer Functions• An attempt made to alleviate mismatch in the training andtesting conditions.• Proposed a new feature called linear predictive ceptrumderived by pole-zero function.26. Speaker Verification Using Mixture DecompositionDiscrimination• Mixture decomposition discrimination (MDD) is based onthe idea that, when modeling speech using hidden Markovmodels (HMM), different speakers speaking the same wordwould cause different HMM mixture components todominate.
27. Recent Advances in the Automatic Recognition ofAudiovisual Speech
28. Unsupervised Speaker Recognition Based onCompetition Between Self-Organizing Maps• Clustering the speaker from unlabeled and unsegmentedconversation, when no priori knowledge about the identityof the participants is given.29. Speaker Recognition with Polynomial Classifiers• Polynomial –based classifier to achieve high accuracy at lowcomplexity.- It has several advantages.1. Polynomial classifier scoring yields a system which is highlycomputationally scalable with the number of speakers.2. A new training algorithm is proposed which is discriminative,handles large data sets, and has low memory usage.3. The output of the polynomial classifier is easily incorporatedinto a statistical framework allowing it to be combined withother techniques such as HMM.
30. Automatic Verbal Information Verification forUser AuthenticationAn example of verbal information verification by askingsequential questions.
Issues and challenges…• Robustness• Portability• Adaptation• Language modeling• Confidence measure• Out of vocabulary words• Prosody
Conclusion…• Problems are still with speaker-generated variability andvariability in channel and recording conditions.• It is very important to investigate feature parameters thatare stable over time, insensitive to the variation ofspeaking manner, including the speaking rate and level,and robust against variations in voice quality due tocauses such as voice disguise or colds.• Studies on ways to automatically extract the speechperiods of each person separately from a dialogueinvolving more than two people have recently appeared asan extension of speaker recognition technology.