Speech Summarization API

The Speech Summarization API transcribes an audio or video recording and summarizes the transcript in one request. It returns the completed transcript, then streams the summary through Server-Sent Events (SSE).

Use this API when your input is a recording. For summaries of existing transcript text or transcript files, use the Summarizer API.

Key features#

• Transcribes and summarizes a single uploaded recording
• Streams Markdown summary text after transcription completes
• Selects a summary template automatically, uses a predefined template, or follows a custom template
• Supports optional citations to transcript evidence
• Configures transcription and summary output languages separately
• Supports speaker diarization and profanity filtering during transcription

How it works#

1. Send a multipart HTTP request with optional settings in config, followed by the recording in file. The config part must come before file.
2. Handle transcribe_started while the API transcribes the recording.
3. Save the transcript from transcribe_complete.
4. Replace the displayed summary with summary_result from each summary_updated event.
5. Treat a final summary_updated event with response_state: "completed", followed by done, as success. Handle an error event or response_state: "error" as a failure.

This API doesn't accept live audio. Upload a complete recording.

For the endpoint, configuration, request example, and event schemas, see Streaming summaries.

Recording requirements#

These limits apply to Speech Summarization uploads. The file-size limit is lower than the Scribe API limit.

Resources#

AI Services are part of the Zoom Build platform. Use Zoom.ai  to create a project, get credentials, and manage your AI Services plan.