v150 · Web API · Speech

Use-Case Sampler

Select a real-world scenario to see which quality tier matches it, how to configure the SpeechRecognition object, and try it live. The right tier balances latency, accuracy, and on-device resource use for each task.

Checking SpeechRecognition support…

🏠

Smart Home Control

Short, single-phrase commands — "turn off lights", "set timer for 5 minutes". Fast response matters more than nuance.

quality: 'command'

low latency · low resource

📝

Note Taking

Long-form prose dictation. Single speaker, moderate noise. Accuracy over speed — you need every word right.

quality: 'dictation'

continuous · interim results

🎙️

Meeting Transcript

Multiple speakers, background noise, complex vocabulary. Highest accuracy tier — tolerates crosstalk and ambient sound.

quality: 'conversation'

multi-speaker · high-noise

Configuration

quality—

continuous—

interimResults—

processLocallytrue (requires local pack)

Select a use case above to begin

Transcript will appear here…

configuration code

Select a use case above to see its configuration code.

implementation reference

Need the exact API surface, compatibility boundaries, errors, lifecycle, and source links? Read the matching gendn reference ↗