Skip to Content
This documentation is provided with the HEAT environment and is relevant for this HEAT instance only.
RunnersAudio Utilsvoice-analysis

voice-analysis

The voice-analysis transform aggregates upstream transcript, audio-metrics, and PCM inputs into a single JSON analysis object (text and per-MD5 voice features).

When to use it

Use at the end of an audio pipeline after generate-transcript when you need a combined report for storage, APIs, or downstream JSON consumers (not Next dashboards).

Limitations

  • Not a dashboard publisher; output is a JSON artifact only.
  • audioAnalysis requires audio-metrics JSON and matching PCM blobs from convert-to-pcm (via collect_pcm_audio).
  • textAnalysis requires transcript JSON from an upstream generate-transcript parent; the parent node instance name in the session template must match the key the processor reads (see the shipped sample voice-analysis-test-r001.json in the audio-utils runner package).
  • Processor docstring marks this node as evolving (placeholder note in output); do not treat fields as a frozen public API yet.

Configuration (summary)

PropertyDefaultDescription
dataSourceName(runner default)Object store for JSON output
textAnalysistrueRun transcript text analysis when transcript input is present
audioAnalysistrueRun per-file audio feature extraction when metrics and PCM exist

Manifest schema lists dataSourceName only; runtime also accepts textAnalysis and audioAnalysis.

Output shape (summary)

{
  "voiceAnalysisVersion": 1,
  "meta": { "generatedAt": "..." },
  "analysis": {
    "text": { },
    "voice": { "<md5>": { } }
  }
}
generate-transcriptCore Utils Runner