Thursday, September 19, 2013



While a forensic expert has the skills to properly match a recording to the person, computers still have a long way to go. Existing voice identification software programs continue to deal with several problems that hinder it from recognizing the speaker, making it ineffectual as a biometric tool. Here are a few of those obstacles.

Stress

Under extreme duress, a person's speech patterns change. His pitch rises, his intonation changes, and his volume increases. As he gets more frustrated or angry, all three factors escalate higher. Voice identification is especially difficult if his usual intervals between words have changed. This is why it's important to be calm and clear when using voice-activated commands on a mobile phone.

Illness

A cold or cough can dramatically change how a narrator sounds. Even voice identification experts will have to exert extra effort to recognize somebody whose stuffy nose hinders them from properly pronouncing their n's and m's. A scratchy throat from excessive coughing will fool voice identification software in the same way. Vault owners who are easily susceptible to the flu should always have an equally secure backup routine in place should their primary speech recognition systems fail.

External Noise

Forget about asking for help along a busy street if the toll-free number directs you to an interactive voice response system. Your words are likely to get caught in between the honking cars and revving engines. The good news is technological breakthroughs in voice identification are now enabling the software programs to separate out different sound waves to isolate human speech.

Developing Vocals

As a person ages, his vocal chords change. One concrete example can be attributed to puberty wherein a boy starts out with a high-pitched voice that becomes an almost inaudible squeak before going down to a baritone level. This makes it challenging even for voice identification experts to recognize a 20-year-old man's voice ten years ago.

Recording Quality

Background din aside, the reverberations from a recording can affect the quality of voice identification as well. These echoes are caused by the sounds bouncing off nearby surfaces. This phenomenon is what makes recording studios particular to acoustics and noise-cancelling microphones.

In no way are voice identification software programs and biometric systems on a par with the real forensic experts. As an emerging technology, the former has made several achievements in the past few years but they are still a long way from being considered a total success.

0 comments:

Post a Comment

Powered by Blogger.

Popular Posts