n8nflow.net logo

Audio to Executive Brief: Transcribe & Analyze with AssemblyAI and GPT-4

by Sridevi Edupuganti•Updated: Last update 14 days ago•Source: n8n.io

Getting Started

🎙️ Audio-to-Insights Workflow (Form Upload + Google Drive Link)

Description

This workflow enables seamless speech-to-text transcription, AI-powered summarization, sentiment analysis, and automated email delivery.
It supports two different input modes:

  • Form Upload (Local File)
  • Form Submission (Google Drive Link)

How it Works

  1. Input

    • Form 1: Upload an audio file (e.g., .mp3,.wav,.mp4)
    • Form 2: Submit a Google Drive link

  2. File Handling

    • Local uploads go directly to AssemblyAI.
    • Drive links are parsed → File ID extracted → File fetched → Sent to AssemblyAI.

  3. Transcription

    • AssemblyAI generates transcript text with punctuation and highlights.
    • Workflow loops with Wait + If until transcript status = completed.

  4. AI Analysis

    • Transcript is passed to OpenAI.
    • Generates a structured output including:
      • Executive summary
      • Sentiment label & score
      • Key points
      • Action items
      • Notable quotes
      • Topics

  5. Email Delivery

    • A formatted email is sent via Gmail with the summary and insights.

Features

  • Dual input support: Google Drive OR direct upload
  • Handles long-running jobs with Wait + If polling
  • AI-powered transcript analysis with structured JSON
  • Automated sentiment scoring and summary generation
  • Professional HTML email reports

Requirements

  • AssemblyAI API Key – transcription
  • Google Drive OAuth2 – file fetch
  • OpenAI API Key – summarization & sentiment analysis
  • Gmail OAuth2 – email delivery

How to Use

  1. Import this workflow into your n8n instance.
  2. Add and configure the required credentials.
  3. Update placeholders for:
    • AssemblyAI API Key
    • Google Drive Link
    • Gmail ID
  4. Trigger via either form (local file or Google Drive link).
  5. 5.For long recordings, split before uploading (10–20 min per chunk, 2–5s overlap).Keep audio consistent (e.g., WAV/MP3, 16 kHz mono if possible).Process chunks sequentially and combine summaries/action items at the end.

Customising this Workflow

  • Adjust the OpenAI prompt to fit your reporting style (executive summary, bullet points, etc.).
  • Extend email formatting with logos or branding.
  • Add Slack, CRM, or Notion integrations for distribution.

Use Cases

  • Meeting or lecture transcription with summaries
  • Podcast analysis with highlights and quotes
  • Business call reviews with action item extraction
  • Academic seminar notes emailed automatically