talk

Overview

Captures voice notes, transcribes them with OpenAI Whisper, optionally labels speakers, and appends both transcript and summary to a Markdown log. The command can limit the number of listening windows and supports summary generation via an agent model.

Usage

ouro talk [--max-questions <n>] <output-path.md>

Key Options

• --max-questions <n> – Cap the number of recording windows (each window runs for the configured chunk duration).

Inputs & Outputs

• Inputs: live audio captured through the configured recorder; environment variables such as OURO_TALK_WHISPER_MODEL, OURO_TALK_SUMMARY_MODEL, and OURO_TALK_CHUNK_SECONDS can fine-tune behaviour.
• Outputs: Appends to (or creates) the target Markdown file with transcript bullet points, optional speaker names, and an agent-generated summary.

When to Use

• Document stakeholder conversations without leaving the terminal.
• Build a running meeting log that feeds requirements or docs flows.

Follow-up Checklist

• Remove temporary audio snippets if they contain sensitive material (the command already attempts cleanup).
• Feed the resulting notes into ouro requirements refine when they contain actionable scope changes.