Automatic Speaker Verification System - Essay Example

In the training step, the characteristics will be derived from a recognized speaker sound then stored in a database that has a model or an outline like Password or Name. The stored information is then extracted by means of the model given by an unidentified speaker. The recovered data is then compared with characteristics extracted from the unidentified speaker sound. If the evaluation result is positive, then the unidentified speaker will be validated.The figure above 3.2.2 demonstrates two things: how the word will be accepted from the speaker's voice as well as how to distinguish the personality of the speaker.

The speech identification will need from the speaker to record the word. The classifier will be employed so as to extract the characteristics of the voice of the speaker and then the representation will be made and saved in the catalog. Just for the speaker identification the speaker's voice will be recorded then features will be taken out and will be evaluated against the other speaker's models so as to accept or reject the personality of a speaker.The figure above 3.2.3 demonstrates the enrollment along with verification phases, in the previous stage, the speaker's voice will be recorded so as to extract the attribute from his/her voice.

After that, the replica will be created for every spoken word and will be stored in the database. During the verification stage, the speaker’s voice will be recorded, and then the traits will be extorted from the speaker's voice. After that, the choice will be made so as to accept or reject the speaker depending on the assessment of the extracted characteristic from a speaker along with the other recognized speaker models that were stored in the database. For this project, I employed the Hidden Markov Model (HMM) for the recognition of speech.3.1 Hardware depiction: The software will be executed by means of some devices such as a Microphone in addition to the computer, the microphone should be precise devoid of the noise in order to record the exact voice and should be selected with some attributes like the outer sponge, alternatively, There is no detailed requirement on the computer that will be employed any computer will be suitable.