Automation: Build a Voice-Note-to-Report-Draft Pipeline
For Patrol Officers ·
What This Builds
An automated pipeline that takes a voice memo from your phone, transcribes it using AI, converts the transcript into a formal police report draft, and emails the draft to your inbox — waiting for you when you return to the station. Instead of transcribing notes and then typing reports, you narrate once and a formatted draft appears in your email.
Prerequisites
- Zapier account (free tier works to start; zapier.com)
- OpenAI account with API access (platform.openai.com — pay-as-you-go, pennies per report)
- Gmail or a personal email account for receiving drafts
- iPhone or Android with a voice memo app
- Comfortable with Otter.ai voice notes (Level 3) or iOS Voice Memos
The Concept
Zapier is like a digital assembly line between apps — when one thing happens (you upload a voice memo to Google Drive), it automatically triggers the next thing (Whisper transcribes it), then the next (ChatGPT converts it to a report), then sends the result somewhere useful (your email inbox). You set up the assembly line once; after that, it runs automatically every time.
Build It Step by Step
Part 1: Set Up Your Trigger — Google Drive Voice Upload
You need a reliable way to get your voice memos into Zapier. The simplest is Google Drive.
- Create a Google account if you don't have one (personal, not department)
- Install the Google Drive app on your phone
- Create a folder in Google Drive called "Field Notes" (or "Patrol Reports")
- On your phone, configure your voice memo app to share recordings to this folder:
- iOS Voice Memos: Record → tap the three dots → Share → Google Drive → select "Field Notes" folder
- Otter.ai: Export → Google Drive → "Field Notes" folder
Part 2: Set Up Your Zapier Account
- Go to zapier.com → Sign up for a free account
- After login, click Create Zap (a "Zap" is one automated workflow)
Part 3: Configure the Trigger Step
- Search for and select Google Drive as your trigger app
- Choose trigger event: New File in Folder
- Connect your Google account when prompted
- Select your "Field Notes" folder
- Test the trigger by uploading a test audio file
Part 4: Add the Transcription Step (OpenAI Whisper)
- Click the + button to add a new step
- Search for OpenAI → choose Create Transcription (Whisper)
- Connect your OpenAI account (you need an API key from platform.openai.com → API keys)
- In the File field, select the file from your Google Drive trigger step
- Set Language to "en" (English)
Cost note: Whisper API costs approximately $0.006 per minute of audio — a 5-minute field narration costs about 3 cents.
Part 5: Add the Report Generation Step (ChatGPT)
- Click + to add another step
- Search for OpenAI → choose Send Message (or Create Chat Completion)
- In the model field, select {{tool:ChatGPT.model:general}} or gpt-4o
- In the User Message field, type:
You are a police incident report writing assistant. Convert these voice-narrated field notes into a formal police incident report narrative. Use third person, past tense, and professional law enforcement terminology. Structure chronologically: arrival, observations, actions, outcome. Flag any missing facts as [NEEDS INFO: ...].
Field notes transcript: [select the transcription text from the Whisper step]
- The [select the transcription text] part is done by clicking the + icon in the field and selecting the output from your Whisper step
Part 6: Send the Draft to Your Email
- Click + to add a final step
- Search for Gmail (or Email by Zapier for non-Gmail)
- Choose Send Email
- Configure:
- To: your personal email address
- Subject: "Report Draft — [add file name from Google Drive step]"
- Body: The ChatGPT output from the previous step
- Turn on your Zap (toggle at the top)
Part 7: Test the Full Pipeline
- Record a test narration on your phone: "Test report. Officer responded to a noise complaint at 123 Main Street at 2200 hours. Officer spoke to [name], who agreed to lower the music. No arrest, civil matter."
- Upload the audio to your Google Drive "Field Notes" folder
- Wait 2-3 minutes
- Check your email — you should receive a formatted report draft
Real Example: End-to-End Flow
Setup: Officer Rivera has the Zap running. After clearing a vehicle burglary scene, while still in the patrol car, she records a 4-minute narration of everything that happened.
Input: She uploads the .m4a audio file to the "Field Notes" folder in her Google Drive app.
What happens automatically:
- Zapier detects the new file in Google Drive
- Whisper API transcribes the 4-minute audio to text
- ChatGPT converts the transcript to a formal vehicle burglary report narrative
- An email arrives in Rivera's inbox with the subject "Report Draft — voice-memo-11-14.m4a"
Output: A 3-paragraph formal report narrative waiting in her email when she returns to the station. She reviews it (2 minutes), copies it to her RMS, verifies all facts, and files the report. Total report time: 4 minutes narration + 2 minutes review = 6 minutes, down from 35 minutes.
Time saved: ~30 minutes per complex incident report.
What to Do When It Breaks
- No email received → Check Zapier task history (left sidebar → Zap History) to see which step failed; the error message will tell you what went wrong
- Whisper transcript has errors → Speak more slowly and clearly; reduce background noise; spell proper nouns; Whisper handles everyday speech well but struggles with police radio chatter in background
- ChatGPT output is too generic → Make your system prompt more specific; add examples of what a good report looks like in the prompt
- OpenAI API key not working → Make sure you have billing enabled at platform.openai.com/billing; new accounts may need to add a payment method before the API activates
Variations
- Simpler version: Skip the Zapier automation entirely — manually paste Otter.ai transcripts into ChatGPT (Level 3 approach). Same result, more manual steps.
- Extended version: Add a step to automatically upload the draft to Google Docs or OneDrive so you have a searchable archive of all your report drafts separate from your RMS
What to Do Next
- This week: Test the full pipeline with 3 different report types; verify accuracy across narration styles
- This month: Refine the ChatGPT prompt in your Zap based on what's missing from early drafts
- Advanced: Add a Google Sheets logging step that records date, file name, and word count for every report draft — gives you a personal productivity log
Advanced guide for patrol officer professionals. This automation uses personal accounts and personal devices — do not process classified case information through personal cloud services. Use for report drafting only; final reports must be verified and filed through official department systems.