Learning linearly separable features for speech recognition using convolutional neural networks
This work addresses speech recognition for automated systems, offering a data-driven approach that is incremental in its use of linear classifiers within a CNN framework.
The paper tackled the problem of speech recognition by learning linearly separable features directly from raw speech using convolutional neural networks, achieving similar or better performance than MLP-based systems with cepstral features.
Automatic speech recognition systems usually rely on spectral-based features, such as MFCC of PLP. These features are extracted based on prior knowledge such as, speech perception or/and speech production. Recently, convolutional neural networks have been shown to be able to estimate phoneme conditional probabilities in a completely data-driven manner, i.e. using directly temporal raw speech signal as input. This system was shown to yield similar or better performance than HMM/ANN based system on phoneme recognition task and on large scale continuous speech recognition task, using less parameters. Motivated by these studies, we investigate the use of simple linear classifier in the CNN-based framework. Thus, the network learns linearly separable features from raw speech. We show that such system yields similar or better performance than MLP based system using cepstral-based features as input.