Transcribe and Summarize a Recorded Community Listening Session

1. Go to otter.ai and sign up with your work email or a Google account.
2. Confirm your email if prompted.
3. Open the app on your phone or the website on a laptop. Either works for recording; the phone app is easier to carry into a community room.

What you should see: A dashboard, mostly empty, with an option to start a new recording. Troubleshooting: If your organization blocks new account signups on the work network, try the signup from your phone's cellular connection instead, then log into the same account on your laptop later.

The free Basic plan includes 300 transcription minutes a month, but caps each individual recording at 30 minutes. A 60-90 minute listening session will hit that wall mid-conversation.

You have two honest options:

• Split the recording. Stop the recording around the 25-minute mark, note where you left off, and start a fresh recording immediately. Otter keeps both as separate files that you can review back to back. This costs you nothing but a short pause and a few seconds of overlap to keep track of.
• Upgrade to Pro ($16.99/user/month), which raises the per-recording cap to 90 minutes. Worth it if your organization runs listening sessions regularly, less worth it for a single one-off session.

Pick one before the session starts so you're not troubleshooting live in front of the room.

Before you press record, say out loud (and ideally have it in any written agenda or sign-in sheet) that the session is being recorded to help capture what's said accurately, that only themes will go into any public report, and that anyone who prefers not to be recorded can still speak and you'll take written notes instead. This isn't just courtesy: recording someone without their knowledge is illegal in some states, and community trust is the whole point of the session.

What you should see: Nods, or a hand raised by someone who'd rather not be recorded. Either is a fine outcome.

1. Open Otter.ai and look for the record button on the home screen. On mobile it's usually a large circular button; on the web dashboard it's typically labeled "Record" or sits behind a plus icon for starting something new.
2. Name the recording something you'll recognize later, like "Riverside Listening Session, [date]."
3. Set the phone or laptop near the center of the room, not off in a corner, so quieter voices still pick up.

What you should see: A live, scrolling transcript appearing on screen as people talk. It won't be perfect in real time. That's fine, it cleans up after processing. Troubleshooting: If the transcript is mostly blank after a minute of conversation, check that the app has microphone permission (a common phone settings issue) and that the device isn't in silent or do-not-disturb mode muting the mic.

If you chose the split-recording route in Step 2, watch the clock. Around minute 25, say something like "let's pause here for a second" out loud (it'll show up in the transcript and helps you find the seam later), stop the recording, and start a new one within a few seconds. Rename the second file to match the first, adding "part 2."

1. When the session ends, tap or click the stop/end button.
2. Otter processes the audio into a cleaned-up transcript and an AI-generated summary. This typically takes one to two minutes after a session of this length, sometimes longer for a full 60-90 minute recording.
3. Open the finished recording from your dashboard.

What you should see: A transcript with speaker labels (Speaker 1, Speaker 2, and so on), plus a short auto-generated summary and a list of suggested action items above it. Troubleshooting: Otter frequently mislabels or fails to name speakers correctly in a group setting with more than two or three people. Don't rely on the speaker labels for attribution. Use the transcript for content and themes, not for "who said what."

Read the auto-generated summary and action items, then skim the full transcript underneath. Otter's summaries have been known to include action items or commitments that were never actually said in the recording. This matters more here than in an ordinary meeting: a listening session summary can end up in a grant report or program plan, and a fabricated "the community requested X" claim is the kind of error that undermines the whole point of asking people what they need. If a claim in the summary doesn't match something you can find in the transcript, cut it or correct it before you use the summary anywhere.

Otter's summary is a starting point, not the finished product. Copy the transcript (or the relevant sections) into ChatGPT or Claude and ask it to group comments into themes, the way you would for a written needs assessment. This step is where the real analytical value shows up, since Otter itself is built for meeting notes, not thematic coding.

Example prompt to paste into ChatGPT or Claude alongside the transcript:

Here's a transcript of a 75-minute community listening session about after-school program needs in our neighborhood. Group the comments into 3-5 themes, note roughly how many people raised each theme, and flag anything that sounds like an urgent or unmet need. Do not invent quotes; only use what's actually in the transcript.