---
title: "Best Practices for voice capture"
source: https://fieldbw.com/docs/sr-7/best-practices
---

# Best Practices for voice capture

## Record First, Process Later

The most effective way to use [Signal Recorder SR-7](/sr-7) is to separate capture from processing. Record freely — don't worry about structure, titles, or organisation in the moment. Let SR-7 handle transcription and summarisation after the fact. Come back later with fresh eyes.

## Use Projects for Context

Group related recordings into projects. A project might be a client engagement, a research topic, a journal, or a creative brief. Projects give structure to your library without forcing you to categorise in the moment — drag recordings into projects whenever it makes sense.

## Star What Matters

Use stars to mark recordings you want to return to. Stars cut across projects and give you a quick-access layer for your most important captures.

## Let AI Do the Filing

AI-generated titles and summaries turn raw recordings into something skimmable. Instead of manually naming recordings, let SR-7 generate titles and then rename only the ones where the AI missed the mark.

## Markdown Export for Your PKM

Enable auto-export to get Markdown files with YAML front matter for every recording. These integrate naturally with Obsidian, Logseq, or any Markdown-based system. The exported files include the transcript, summary, and metadata — making your recordings part of your knowledge graph.

## Hardware Recorders

If you use a dedicated recorder like the teenage engineering TP-7, make SR-7 your import hub. Record on hardware for the tactile experience, then import into SR-7 for transcription, search, and organisation. See the [hardware import guide](/docs/sr-7/importing-from-hardware) for setup details.