Category · 5 terms
Natural Language Processing
defined clearly.
Techniques and models for understanding and generating human language.
Automatic Speech Recognition (ASR)
Converting spoken audio into text — the foundation of voice assistants, transcription services, and most speech-to-text workflows.
BERT (Bidirectional Encoder Representations from Transformers)
Google's 2018 encoder-only transformer that revolutionised NLP by pretraining on masked-token prediction in both directions.
Byte-Pair Encoding (BPE)
A subword tokenization algorithm that iteratively merges the most frequent pair of adjacent symbols, building up a vocabulary of subword units.
Speaker Diarization
Answering "who spoke when" — segmenting a multi-speaker recording into regions attributed to each distinct speaker, without necessarily knowing who they are.
Tokenization
The process of converting raw text into tokens — discrete units that a language model can consume. Dominated by subword methods (BPE, WordPiece, SentencePiece).