Features extraction in ASR
Features extraction is an important preprocessing stage in a DL pipeline of ASR. This stage consists of an analyzer and the extraction of audio fingerprints or features. This stage also mainly computes a sequence of feature vectors, which provides a compact representation of a gathered speech signal. Generally, this task can be performed in three key steps. The first step is known as speech analysis. This step carries out a spectra-temporal analysis of the speech signal and generates raw features describing the envelope of the power spectrum of short speech intervals. The second step extracts an extended feature vector that consists of static and dynamic features. The final step converts these extended feature vectors ...
Become an O’Reilly member and get unlimited access to this title plus top books and audiobooks from O’Reilly and nearly 200 top publishers, thousands of courses curated by job role, 150+ live events each month,
and much more.
Read now
Unlock full access