Speaker independent phoneme recognition by MLP using wavelet features
نویسندگان
چکیده
Feature extraction is one of the most important tasks in speech recognition system. Most of the speech recognition systems use Short Time Fourier Transform (STFT) for the derivation of features from the spoken utterances. In this paper we try to exploit the higher time–frequency resolution property of Discrete Wavelet Transform (DWT) for extraction of speaker independent features. The features are extracted every 8ms to account for the faster changes in the phoneme. These features are then used to train a Multi-Layer Perceptron (MLP) classifier for the recognition of phonemes.
منابع مشابه
Integrating MLP features and discriminative training in data sampling based ensemble acoustic modeling
In this paper, we propose to incorporate the widely used Multiple Layer Perceptron (MLP) features and discriminative training (DT) into our recent data-sampling based ensemble acoustic models to further improve the quality of the individual models as well as the diversity among the models. We also propose applying speaker-model distance based speaker clustering for data sampling to construct en...
متن کاملSpeaker Identification System based on PLP Coefficients and Artificial Neural Network
Speaker recognition is the process of automatically recognizing who is speaking on the basis of individual information included in speech waves. Feature extraction for speech recognition is a subject of a major interest today; different features have been investigated in speech recognition systems. The perceptual linear predictive PLP: this technique uses three concepts from the psychophysics o...
متن کاملAugmenting short-term cepstral features with long-term discriminative features for speaker verification of telephone data
Short-term cepstral features have long been chosen as standard features for speaker recognition thanks to their relevance and effectiveness. In contrast, discriminative features, calculated by a multi-layer perceptron (MLP) from much longer stretches of time, have been gradually adopted in automatic speech recognition (ASR). It has been shown that augmenting short-term cepstral features with lo...
متن کاملNeural Network based Classification for Speaker Identification
Speaker Recognition is a challenging task and is widely used in many speech aided applications. This study proposes a new Neural Network (NN) model for identifying the speaker, based on the acoustic features of a given speech sample extracted by applying wavelet transform on raw signals. Wrapper based feature selection applies dimensionality reduction by kernel PCA and ranking by Info gain. Onl...
متن کاملA Chinese phoneme clustering theory and its application to a text independent speaker verification system
This paper presents a new idea of Chinese phoneme clustering and a text independent speaker verification system with this technique applied. It changes the way of conventional verification method with averaging features used, instead, both the dynamic and static features of speech are included in our new method. Also it leads to fast and efficient clustering algorithm in the training phase. The...
متن کامل