Automate Audio Transcription & Summary with n8n, Google Drive & OpenAI

Opening Problem Statement

Meet Sarah, a busy podcast producer who handles dozens of audio recordings weekly. Each recording needs to be transcribed and summarized before sharing key points with her team. Manually doing this takes her over 4 hours per week, cutting into her time for creative tasks and sometimes causing errors in transcription and summaries.

Imagine the frustration of juggling multiple file uploads, copying content into transcription tools, and manually creating summary documents. This workflow shows how Sarah can automate all these steps by linking Google Drive, OpenAI’s Whisper, and Notion via n8n, drastically reducing manual effort and errors.

What This Automation Does

When a new audio file is uploaded to a specific Google Drive folder, this workflow triggers and runs the following steps automatically:

• Detects the new audio file creation in Google Drive with the Google Drive Trigger node.
• Downloads the new audio file through the Google Drive node for further processing.
• Sends the audio file to OpenAI’s Whisper model to transcribe speech to text.
• Uses OpenAI GPT-4 Turbo to generate a rich JSON summary of the transcript, including title, key points, action items, and sentiment.
• Creates a new page in Notion with the transcription summary for easy team collaboration.
• Automates continuous polling every minute to catch new uploads promptly.

This flow saves Sarah multiple hours per week by eliminating manual transcription and note-taking, ensures greater accuracy with AI, and keeps her team updated seamlessly.

Prerequisites ⚙️

• n8n account with the ability to run workflows.
• Google Drive account authorized for OAuth2 access.
• OpenAI API account with access to Whisper for transcription and GPT-4 Turbo for summarization.
• Notion account with API integration set up to allow page creation.
• Recommended: A self-hosted n8n instance for better control and privacy (see Hostinger guide).

Step-by-Step Guide to Building this Workflow

1. Set up Google Drive Trigger to Monitor Folder
Navigate: Click + New Node → Choose Google Drive Trigger
Configure:
Select Event: fileCreated, mode: everyMinute, and set the folder to watch by its ID (e.g., folder ID “182i8n7kpsac79jf04WLYC4BV8W7E_w4E” for “Recordings” folder).
Outcome: Workflow triggers whenever a new audio file is added.
Common mistake: Forgetting to select the correct folder ID causes trigger failure.

2. Download the Uploaded Audio File
Navigate: Add a Google Drive node.
Configure:
Operation: download.
File ID: Use {{$json.id}}, File Name: {{$json.originalFilename}}.
Outcome: The audio file’s binary content is downloaded for processing.
Common mistake: Not mapping the file ID from trigger output correctly.

3. Transcribe Audio with OpenAI Whisper
Navigate: Add OpenAI node (langchain category).
Configure:
Resource: audio, Operation: transcribe.
Provide audio file binary from previous node.
Outcome: Get back plain text transcription.
Common mistake: Not feeding binary content properly into the audio field.

4. Summarize Transcript Using OpenAI GPT-4 Turbo
Navigate: Add second OpenAI node.
Configure:
Model: gpt-4-turbo-preview.
System prompt: A detailed JSON-based summarization instruction that extracts structured data including title, summary, key points, actions with dates, stories, references, arguments, themes, sentiment.
User message: Include the transcript text (from first OpenAI node) using expressions.
Outcome: Receive a structured JSON summary.
Common mistake: Not enclosing transcript within the system context or improper JSON formatting.

5. Create a New Page in Notion with Summary
Navigate: Add Notion node.
Configure:
Page ID where the new page is created.
Title: Extract from summary JSON, for example {{JSON.parse($json.message.content).audioContentSummary.title}}.
Blocks: Add heading 1 “Summary” and the summary text.
Outcome: Notion page is created with readable summary.
Common mistake: Using incorrect JSON path or empty page ID.

Customizations ✏️

• Modify the polling interval on the Google Drive Trigger node from every minute to a different interval under “pollTimes” for less frequent checks.
• Customize OpenAI summarization prompt to add or remove summary fields based on your workflow needs by editing the system content prompt in the second OpenAI node.
• Add additional nodes after Notion to notify your team via Slack or email by connecting Slack or Gmail nodes.
• Include metadata or tags in the Notion page from transcription for better classification using extra blocks.
• Extend error handling by adding IF or Error Trigger nodes to manage transcription failures or API limits gracefully.

Troubleshooting 🔧

Problem: “Google Drive Trigger does not start the workflow when new files are added.”
Cause: Incorrect folder ID or missing OAuth2 permissions.
Solution: Verify and reselect the correct folder ID in the trigger node. Check your Google Drive API permissions in the linked OAuth2 credentials.

Problem: “OpenAI transcription node returns an error or empty response.”
Cause: Audio file binary not correctly passed or unsupported file format.
Solution: Ensure the binary property name matches and audio file is in a supported format like mp3 or wav.

Problem: “Notion page creation fails with invalid page ID or permission error.”
Cause: Incorrect Notion integration setup or missing page ID.
Solution: Confirm Notion API token has write permissions and check the page ID or use a database if needed.

Pre-Production Checklist ✅

• Verify Google Drive OAuth2 connection works with correct folder access.
• Test uploading a sample audio file to trigger the workflow.
• Ensure OpenAI nodes are connected with valid API keys for transcription and GPT-4 Turbo.
• Confirm Notion API integration allows creating pages.
• Review all JSON expression mappings between nodes to prevent runtime errors.

Deployment Guide

Activate the workflow in n8n by toggling the active switch.
Monitor initial runs for any errors and review logs to troubleshoot issues.
Setup notification nodes to alert you on errors or completion if desired.
For scaling, consider usage limits on API or splitting uploads across multiple workflows.

FAQs

Q: Can I use Dropbox instead of Google Drive?
A: Yes, replacing the Google Drive trigger with a Dropbox trigger node is possible but requires changing the node configurations accordingly.

Q: Does OpenAI transcription use up API quota?
A: Yes, each transcription request consumes API tokens, so monitor usage to stay within your plan limits.

Q: Is my audio data secure?
A: Data security depends on your OpenAI and Google Drive account policies. Using self-hosted n8n enhances control.

Q: Can this handle long audio files?
A: Whisper can process fairly long audio, but very large files might need splitting or chunking.

Conclusion

By following this detailed guide, you have automated the tedious process of transcribing and summarizing audio recordings, a massive time saver for roles like podcast producers, journalists, or content creators like Sarah. This workflow leverages Google Drive’s file creation trigger, OpenAI’s powerful Whisper transcription and GPT-4 Turbo summarization, and Notion for structured note keeping.

You’ve cut down hours of manual work, improved accuracy, and streamlined team collaboration. Next, you might explore adding multi-language support, integrating with task management tools, or alerting your team via Slack or email. Start saving time and making your audio content more accessible today!