Comments on: Another Biometric Marker… http://www.blindside.org.uk/2007/07/24/another-biometric-marker/ What's going to go wrong in our e-enabled world? Sat, 22 Nov 2008 11:30:38 +0000 http://wordpress.org/?v=2.3.1 By: William Heath http://www.blindside.org.uk/2007/07/24/another-biometric-marker/#comment-788 William Heath Tue, 31 Jul 2007 11:16:14 +0000 http://www.blindside.org.uk/2007/07/24/another-biometric-marker/#comment-788 Cheers Phil. Hang on; I'm perplexed. Are you the Phil people inside government are too scared to talk to because they seem to think you're a raving anarchist with extreme views and even a ponytail? Or are you the Phil I know as a perfectly respectable generally law-abiding taxpayer with a lot of useful expert knowledge and a cracking sense of humour and some public sculptures on display in Wales? Are these duplicate identities or are you one and the same person? Hang on; I'd better recalibrate my Weltauffassung...I think my head is about to explode :-) Cheers Phil. Hang on; I’m perplexed. Are you the Phil people inside government are too scared to talk to because they seem to think you’re a raving anarchist with extreme views and even a ponytail? Or are you the Phil I know as a perfectly respectable generally law-abiding taxpayer with a lot of useful expert knowledge and a cracking sense of humour and some public sculptures on display in Wales? Are these duplicate identities or are you one and the same person? Hang on; I’d better recalibrate my Weltauffassung…I think my head is about to explode :-)

]]> By: Phil Booth http://www.blindside.org.uk/2007/07/24/another-biometric-marker/#comment-690 Phil Booth Tue, 24 Jul 2007 20:45:37 +0000 http://www.blindside.org.uk/2007/07/24/another-biometric-marker/#comment-690 Voiceprint matching is an entirely different problem from voice recognition. The former is about one-to-one matching of the characteristics of a captured soundwave with a physical model of an individual's biological voice apparatus (which is very hard to 'forge' and therefore potentially a useful biometric), whereas voice recognition is about one-to-many matching of many people's spoken version of a word versus a 'generic' version of that word. You still have to train even quite good voice recognition systems to get them to perform well on broad vocabularies, but there are systems that aren't too bad at recognising limited sets of words (e.g. numbers spoken over the phone). Each (one hopes) is fit to the specific purpose for which it was designed... Voiceprint matching is an entirely different problem from voice recognition. The former is about one-to-one matching of the characteristics of a captured soundwave with a physical model of an individual’s biological voice apparatus (which is very hard to ‘forge’ and therefore potentially a useful biometric), whereas voice recognition is about one-to-many matching of many people’s spoken version of a word versus a ‘generic’ version of that word.

You still have to train even quite good voice recognition systems to get them to perform well on broad vocabularies, but there are systems that aren’t too bad at recognising limited sets of words (e.g. numbers spoken over the phone). Each (one hopes) is fit to the specific purpose for which it was designed…

]]>