Voice2Note: I Built a Local AI Meeting Assistant for Freelancers

Description and use

Voice2Note automates transcription, structured extraction, and follow‑up email drafting to reduce administrative work for freelancers.

Problem

Freelancers need to review meeting recordings, extract decisions, tasks, deadlines, and draft follow‑up emails, but manual review is time‑consuming and risks missed commitments.

Outcome

Voice2Note automates transcription, structured extraction, deadline classification, email drafting, and semantic search, enabling freelancers to reduce administrative overhead and avoid missed commitments.

Workflow

• Record or import voice recording
• Local Whisper model transcribes audio to text
• Llama 3.2 3B via Ollama extracts structured information (summary, decisions, tasks, commitments, deadlines)
• Zod validates extracted structure
• Store transcript and embeddings in SQLite
• Index content for semantic search
• Generate editable follow‑up email draft
• Export tasks as CSV and summaries as Markdown

Evidence

Voice2Note turns that workflow into a structured process.

Source

Record or import voice recordings and generate transcripts.

Source

Local Whisper model transcribes audio to text

Source

Llama 3.2 3B via Ollama for structured extraction

Source

The application is designed to help freelancers reduce repetitive administrative work while retaining control over confidential client information.

Source