Experience · Software component

Streaming Speech Recognizer

Software componentExperienceExperience & Human Oversightarc:StreamingSpeechRecognizer

A speech recognizer that processes live audio in short chunks and returns provisional partial transcripts immediately, marking segments final once enough context has arrived.

Responsibility. Delivers low-latency incremental transcripts of live speech.

Also known as: Streaming ASR, Real-time recognition

Variant of Automatic Speech Recognizer abstract

When to choose. Choose when real-time interaction is required (live customer-service calls, voice assistants, live accessibility captioning) and latency matters more than perfect accuracy.

Direct neighbourhood (hover for relationship types)

Relationships

sends data to dynamic

Voice Turn Coordinator Ch7.5

alternative to variability

Speech Transcriber Ch7.5

Design guidance

MUST treat intermediate transcripts as provisional; consumers SHOULD either wait for finalized segments or process incrementally with correction.

Quantitative guidance

As stated by the sources; verify before use.

Processes audio in 80-160ms chunks (Ch7.5).

Classification

Patterns: Incremental partial results with is_final flagEarly-context processing
Technologies: NVIDIA Riva ASR (streaming mode)
Quality attributes: Performance efficiency (ISO/IEC 25010)
Risks mitigated: Waiting for silence before agent processing begins

Sources

Ch7.5: T. Nguyen, "NeMo Curator, Riva Speech AI & Multimodal Integration," in Mastering Agentic AI Systems: Guide for the NVIDIA NCP-AAI Exam, 1st ed. 2026, ch. 7.5. ISBN: 9798244538229.