Rapport · Engineering + Product case study
From hours of audio to a promise someone can keep.
Continuous recordings hide useful moments among unrelated speech. I built a path from audio to conversations, then to findings linked to the words behind them.
PythonAudio pipeline
PydanticOutput validation
The decisions
Three decisions, the reasoning behind each, and what changed as a result.
Hours of audio. Minutes you can trace.
Chunking makes processing manageable. Original timestamps keep the output connected to the recording.
The output remains traceable to a moment, not just a large transcript.
How it works
A quick guide to the interactive flow on this page.
Dynamic contentClick each step to see a different part of the story.
The outcome
A finding that leads back to the conversation.
Promises and customer risks become actionable views for the team, with speaker context and timestamps preserved. The example here illustrates the pipeline’s output.