Speech Communication Process

1527 Words7 Pages
INTRODUCTION TO SPEAKER RECOGNITION
1.1 INTRODUCTION
Speech communication is a natural phenomenon among human beings. The intended message is transferred from one person to another through the complex mechanisms of speech production and speech perception. Speech production begins when the intended message represented in some abstract form in the mind of the speaker is converted into neural signals. These neural signals control the human vocal system to produce an acoustic wave. This acoustic wave is successfully decoded by the speech perception mechanism of the listener to realize the intended message.
Fig
The speech production mechanism is understood better by studying the anatomical structure of the human vocal system shown in Fig.1.1. The
…show more content…
1.4 ISSUES IN SPEAKER RECOGNITION
Speaker recognition by a machine incorporates three stages. They are, (1) abstraction of features to represent the speaker information being in the speech signal, (2) probabilistic modeling of elongated speaker features, and (3) decision logic to develop for the identification or verification task. The issues involved in each of these stages are discussed below.
1.4.1 Extraction of Speaker Information
The fundamental task in a speaker recognition system is to abstract features from the speech signal to represent the speaker information. It is known that human beings use, powerful features like speaker's vocabulary, style of speech and verbal habit to recognize speakers. Intuitively, it is clear that these features constitute important speaker information. The difficulty arises in representing these features due to limitations of the existing feature extraction techniques [1]. Current speaker recognition systems use segmental features such as vocal tract shape to represent the speaker-specific information. These features show significant variations across speakers, but they also show considerable variations from time to time for a single
…show more content…
The unfamiliar speaker is identified as the speaker whose model matches best with the given utterance. The use of such a simple logic leads to: (1) increase in the identification time with an increase in the population of registered speakers, and (2) possibility of an unfamiliar speaker being identified as one of the registered speakers. The given speech utterance is related to the model of the speaker whose identity is claimed is the task of speaker verification. It may accept or reject the identity claim based on the threshold comparison between the model and the given

More about Speech Communication Process

Open Document