Speech recognition

Speech recognition is the computational conversion of spoken language into written text or structured output. It infers linguistic content from an audio signal, often despite variation in speakers, accents, and recording conditions.

Automatic speech recognition: The use of computational systems to convert speech audio into text or other representations. The broader technical task includes the transcription and structured-output systems built from speech recognition.
Phonetics: The study of the physical production, transmission, and perception of speech sounds. Recognition depends on how spoken sounds are produced and represented acoustically.
Voice user interface: An interface that lets people control a system or access information through spoken interaction. Speech recognition turns spoken commands into inputs that a voice interface can act on.
Speech synthesis: The generation of spoken audio from text or other linguistic representations. It runs in the opposite direction: recognition maps speech to output, while synthesis produces speech.
Acoustic model: A model that relates speech sounds or audio features to phonetic or linguistic units. It connects the measured signal to candidate speech units.
Phoneme: A contrastive sound category in a language that can distinguish one word from another. Many systems map audio to phonemes or related subword units before forming words.
Dictation software: Software that converts spoken language into text for composition or data entry. It uses recognition to produce editable text from continuous speech.
Speaker recognition: The identification or verification of a person from characteristics of their voice. It identifies who is speaking rather than what the speaker said.
Language model: A model that assigns probabilities to sequences of linguistic units, such as words or tokens. It favors word sequences that are plausible in the language, resolving acoustically similar alternatives.
Spectrogram: A visual representation of signal energy across time and frequency. It reveals the changing acoustic patterns from which speech features are extracted.

Linked from 28 pages