Glossary → Transcription & speech recognition
Transcription & speech recognition

Beam search

Beam search is a decoding strategy where a speech-recognition model keeps several candidate transcriptions in play as it processes audio, choosing the overall most probable one at the end instead of committing word by word.

The alternative, greedy decoding, always takes the single most likely next token — faster but prone to locking in early mistakes. Beam width trades accuracy against speed and is one of the knobs behind 'fast' vs 'accurate' transcription modes.

Related terms

Greedy decodingGreedy decoding is the fastest way for a speech model to produce text: at every step it emits the single most … Language model (in ASR)In speech recognition, a language model supplies knowledge of which word sequences are likely, letting the sys… Automatic speech recognition (ASR)Automatic speech recognition (ASR) is the technology that converts spoken language in audio into machine-reada…

Frequently asked

What is Beam search?

Beam search is a decoding strategy where a speech-recognition model keeps several candidate transcriptions in play as it processes audio, choosing the overall most probable one at the end instead of committing word by word.

Why does Beam search matter?

The alternative, greedy decoding, always takes the single most likely next token — faster but prone to locking in early mistakes. Beam width trades accuracy against speed and is one of the knobs behind 'fast' vs 'accurate' transcription modes.

What terms are related to Beam search?

Closely related concepts: Greedy decoding, Language model (in ASR), Automatic speech recognition (ASR) — each has its own entry in this glossary.

Put the term to work

Transcribe audio or video in 99+ languages — speaker labels, word timestamps, captions.

Transcribe a file →