DocsAPI reference

Sentiment analysis

Add conversation-level or segment-level sentiment signals to a speech pipeline, scored from the transcript alongside speakers and entities.

Sentiment analysis adds a derived signal about the language in a transcript. It can support aggregate conversation analytics, review prioritization, and application-specific classification.

Interpret it as a signal

Sentiment is model-derived, context-dependent, and not a direct measurement of a person's internal state. Do not use it as the sole basis for high-impact decisions.

Choose the right scope

  • Segment-level output can show changes across a conversation.
  • Speaker-aware output can keep signals attached to the correct participant.
  • Conversation-level output supports broad categorization but can hide important turns.

Evaluate against language, dialect, domain, and conversational style representative of production traffic.

Entity detection

Extract structured references such as people, organizations, places, dates, and domain-specific values from transcripts.

LLM processing

Chain speech-derived data into hosted language models for extraction, classification, summaries, and custom transformations.

On this page

Interpret it as a signalChoose the right scope