Feature vector that contains the most commonly used features when converting a raw wavefile to a sequence of features.

Hamilton College

The first step in ASR is to transform the input waveform into a sequence of acoustic feature vectors, where each vector represents the information in a small time window of the signal.

Feature Vector

Spectral information is extracted on a mel scale, rather than equally across all frequency bands. We do this because humans hearing is more sensitive at lower frequencies than higher ones, so modeling speech recognition the same way helps performance.

Mel Filter Bank and Log

An on-going but a helpful book resource about NLP
https://web.stanford.edu/~jurafsky/slp3/

Speech and Language Processing (3rd ed. draft) 

Log Mel Spectrum

In feature extraction for ASR, we must extract spectral information from our windowed signal and find out how much energy the signal contains at different frequency bands. This is done with DFT.

Spectral Information Extraction

The input to a Listen, Attend, and Spell (LAS) model is a sequence of $$t$$ acoustic feature vectors $$F = f_1, f_2, \dots, f_t$$, where each vector spans a frame of 10 milliseconds. Assuming the output consists of letters, the output sequence is $$Y = (\langle SOS \rangle, y_1, \dots, y_m, \langle EOS \rangle)$$, where $$\langle SOS \rangle$$ is a special start-of-speech token and $$\langle EOS \rangle$$ is a special end-of-speech token. The associated image shows the character set often selected for the English language.

Learn Before

Related