Introduction to Automated Speech Recognition (ASR)

Automated Speech Recognition is the problem of translating the time-domain acoustic signal to a string of words.

Application

• Human-computer Interaction
• Automatic Transcription

Phonetic Alphabets

Speech Recognition systems generally work with phonetic alphabets (phones). A Phone is a unit of speech.

There are two common Phonetic Alphabet Systems in use:

Internation Phonetic Alphabet (IPA) is a global standard with the goal of transcribing every spoken language in the world.

ARPAbet is an ASCII representation of the North-American English part of IPA.