015. Speech Spectrograms catch criminals

This time I tell you about an exciting new technique, by which we can recognize a person from his speech. This technique is useful in cases like abduction, when kidnapper asks for ransom over the telephone. Can a person be positively recognized from his voice? You must have noticed that you can readily recognize a person from his voice. Even when you are sitting engrossed in your work and someone calls you from the back, you can readily recognize his voice. This is because, the voice-generating system in each individual is different. In fig(1) you can see the various structures of the mouth which are involved in speech production such as palate, tongue, teeth, lips, cheeks, nasal aperture etc. These structures are different in different individuals and this is the basis of speech recognition.

For a long time scientists were thinking of having a kind of "photographs" of human speech. It is always easier and simpler to compare two photographs, than to compare two sounds. Moreover, if a person has not heard the voice of a culprit for a long time, and he was asked to recognize that culprit again by his voice after a period of, say ten years, he might be at a loss to recognize him. If there was no way of converting the sound to a photograph, a police officer would face this problem very often. To overcome this handicap, in 1963, as scientist Lawrence G.Kersta, developed a unique machine in his laboratory in Somerville, New Jersey. This machine is known as the speech spectrograph and is shown Fig(2).

The speech spectrograph converts a human speech into various graphs which can be easily compared. Fig 3, shows the spectrograms (the voice graphs) of male and female voice. Once can easily spot the difference between the two.

Fig 4, shows the spectrograms or voice prints of 5 people saying a single word"you". there are 6 spectrograms which show clearly that one person has said" you" two times.

Try to visualize the photograph carefully and try to find out which two voice prints look similar. These are the words uttered by the same person.

With this background in mind, it is now easier to explain, how the machine helps in police investigation. Suppose a kidnapper has kidnapped a child, and then asks for ransom over the phone. The police scientist can record his voce and make speech spectrograms from that recorded voice, by the help of the spectrograph. The voice prints are then stored in the police prints are then stored in the police records for future reference. In future whenever a suspect is arrested, his voice prints can be made again and matched with that of the telephone caller. If the voice prints match perfectly, then the suspect is indeed the real kidnapper, otherwise not. The same technique can he used in several other situations. Sometimes people phone young females just for fun sake and speak obscene languages over the phone. These are basically harmless people, but their repeated malicious calls may be a nuisance. Till now the only way to tackle with such malicious callers was to keep the phone off the hook or to keep the receiver off the cradle. But now the affected party can call the police and ask them to make voice prints of such callers. These voice prints are permanent records in the files of the police and can be used anytime later when a suspect is arrested.

Several countries are using this new technique now regularly to investigate crime.. Michigan State Police for example have been using this method with success since 1968. Some very famous crimes have been solved by this technique.

This account should give you an idea of how deeply modern science has affected crime investigation. This is a new field of crime investigation. This is a new field of crime investigation and more research is needed before we can say the final word on it.

Note: Certain figures mentioned in the text may not be visible here. Although not being able to look at the figures would not interfere with the comprehension of the article very much, if it is desired to have a look at the figures, readers are advised to consult the original text in the magazine.

IMPORTANT NOTE: THIS MATERIAL IS COPYRIGHTED BY THE AUTHOR AND MAY NOT BE REPOSTED, REPRINTED OR OTHERWISE USED IN ANY MANNER WITHOUT THE WRITTEN PERMISSION OF THE AUTHOR