01751nam a2200217   4500001000700000100004100007101000800048200014300056210002500199215002000224300014900244330099400393606002901387606001801416606001801434606001701452606001301469606000901482606001901491700002301510394577  a20170525d2017       y0pory50      ba  aeng  aA framework for performing forensic and investigatory speaker comparisons using automated methodsbRecurso eletrónicofDavid Brian Marks  aDenverc[s.n.]d2017  a1 CD-ROMd12 cm  aTese de Mestrado apresentada à Universidade do Colorado, tendo como orientadora Catalin Grigoras. Ficheiro de 2,99 MB em formato PDF (174 p.).  aI. INTRODUCTION. Terminology. Challenges of Voice Forensics. Scope. II. BACKGROUND. Scientific Foundations. The Scientific Method. Bias Effects. Legal Foundations. Rules of Evidence. Federal Case Law. State Case Law. Factors in Speaker Recognition. The Nature of the Human Voice. Speaker Recognition Systems. Bias Effects. Standards. Historical Baggage. III. COMPARISON FRAMEWORK. Case Assessment. Forensic Request. Administrative Assessment. Technical Assessment. Decision to Proceed with Analysis. Analysis and Processing. Data Preparation. Data Enhancement. Selection of the Relevant Population. System Performance and Calibration. Combining Results from Multiple Methods or Systems. Conclusions. Interpreting Results. Communicating Results. Case Studies. Case Study 1. Case Study 2. Case Study 3. SUMMARY AND CONCLUSIONS. Challenges in the Relevant Population. Fusion for Multiple Algorithms. Verbal Scale Standards for Reporting Results. Data and Standards for Validation. REFERENCES.  aIDENTIFICAÇÃO DA VOZ  aEXAME À VOZ  aREGISTO VOCAL  aMETODOLOGIAS  aSOFTWARE  aTESE  aESTADOS UNIDOS  aMARKSbDavid Brian