n8n Workflow: Browser Audio Recording Transcription and AI Analysis with Deepgram and GPT-4o

This n8n workflow automates the process of transcribing audio recordings and performing AI-driven analysis on the transcribed text. It's designed to be triggered by a webhook, accept an audio file, transcribe it using an external API (likely Deepgram, based on the directory name hint), and then use an AI model (likely GPT-4o) to analyze the content.

What it does

This workflow streamlines the following steps:

1. Receives Audio via Webhook: It starts by listening for incoming HTTP POST requests, expecting an audio file as part of the payload.
2. Prepares Audio for Transcription: The received audio data is converted into a file format suitable for an external transcription service.
3. Transcribes Audio: An HTTP request is made to an external API (e.g., Deepgram) to transcribe the audio file into text.
4. Structures Transcription Output: The raw transcription output is processed and structured using a "Set" node, likely extracting the plain text transcription.
5. Performs AI Analysis: The transcribed text is then fed into an AI Agent, which utilizes an OpenAI Chat Model (likely GPT-4o) and a Structured Output Parser to analyze the text and extract specific information or insights.
6. Responds to Webhook: Finally, the workflow sends back a response to the initial webhook trigger, likely containing the transcription and the AI analysis results.

Prerequisites/Requirements

To use this workflow, you will need:

• n8n Instance: A running instance of n8n.
• Deepgram API Key (Likely): Although not explicitly defined in the JSON, the workflow's purpose strongly suggests an API key for an audio transcription service like Deepgram for the "HTTP Request" node.
• OpenAI API Key: For the "OpenAI Chat Model" node used by the "AI Agent" for text analysis.
• Google Drive Credentials (Optional, but present in workflow): While not directly connected in the provided JSON, Google Drive nodes are present, suggesting potential for storing or retrieving audio/transcription data. If these are to be used, you'll need Google Drive OAuth credentials.
• Gmail Credentials (Optional, but present in workflow): Similar to Google Drive, Gmail nodes are present. If email notifications or actions are intended, Gmail OAuth credentials will be required.

Setup/Usage

1. Import the Workflow:
   • Save the provided JSON content as a .json file.
   • In your n8n instance, go to "Workflows" and click "New".
   • Click the "Import from JSON" button and select the saved JSON file.
2. Configure Credentials:
   • HTTP Request (Transcription): Locate the "HTTP Request" node (ID: 19). You will need to configure its URL, method, headers (including your Deepgram API key if using Deepgram), and body to send the audio file for transcription.
   • AI Agent / OpenAI Chat Model: Locate the "OpenAI Chat Model" node (ID: 1153). Configure your OpenAI API key credential.
   • Google Drive / Gmail (if used): If you plan to extend the workflow to use the Google Drive or Gmail nodes, configure their respective OAuth 2.0 credentials.
3. Configure Webhook Trigger:
   • Locate the "Webhook" node (ID: 47).
   • Set the "HTTP Method" to POST.
   • Note down the "Webhook URL". This is the endpoint you will send your audio recordings to.
4. Customize AI Agent:
   • Locate the "AI Agent" node (ID: 1119). You will need to define the "System Message" and "User Message" to instruct the AI on how to analyze the transcribed text.
   • The "Structured Output Parser" (ID: 1179) will need a schema defined to guide the AI in returning structured data.
5. Activate the Workflow:
   • Ensure all necessary nodes are configured.
   • Click the "Activate" toggle in the top right corner of the n8n editor to enable the workflow.
6. Send Audio:
   • Send an HTTP POST request containing your audio file (e.g., as binary data or a file upload) to the Webhook URL obtained in step 3. The workflow will then process the audio, transcribe it, analyze it, and respond.