Voice-To-Prompt-Pipeline

Convert audio recordings into structured AI prompts using Gemini 2.5 Pro's multimodal audio capabilities.

Overview

This CLI tool processes audio recordings of AI prompts and transforms them into well-structured markdown documents. It uses Google's Gemini 2.5 Pro model to transcribe and reformat the audio according to a specific template that separates core requests from contextual information.

Features

• Interactive file selection from the in/ directory
• Multimodal audio processing using Gemini 2.5 Pro
• Automatic transcription with filler word removal
• Structured output with summary, requests, and context sections
• Timestamped output files in markdown format

Prerequisites

• Python 3.8 or higher
• uv (recommended) or pip
• Google Gemini API key

Setup

1. Create a virtual environment using uv:

uv venv
source .venv/bin/activate  # On Windows: .venv\Scripts\activate

2. Install dependencies:

uv pip install -r requirements.txt

Alternatively, if not using uv:

python -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt

3. Configure your Gemini API key:

Copy the example environment file and add your API key:

cp .env.example .env

Then edit .env and replace the placeholder with your actual Gemini API key:

GEMINI_API_KEY=your-actual-api-key-here

Usage

1. Activate the virtual environment (if not already activated):

source .venv/bin/activate

2. Place your audio file(s) in the in/ directory

3. Run the CLI:

python process_audio.py

4. Select the audio file you want to process from the interactive menu

5. The structured prompt will be saved to the out/ directory with a timestamp

Supported Audio Formats

• MP3
• WAV
• M4A
• OGG
• FLAC
• AAC
• WMA
• OPUS

Output Format

The generated markdown file follows this structure:

Summary

A two-line summary of the prompt containing core requests and contextual data.

Requests

Organized task list of AI requests.

Context/Background

Organized presentation of context data, which may reference specific tasks.

System Prompt

The tool uses the system prompt defined in system-prompt.md. This ensures Gemini processes the audio according to your specific requirements for formatting and organization.

Example

$ python process_audio.py

==================================================
Voice to Prompt Pipeline
==================================================

System prompt loaded from: system-prompt.md

Available audio files:
--------------------------------------------------
1. 1.mp3 (8.60 MB)
--------------------------------------------------

Select a file (enter number): 1

Processing: 1.mp3
Uploading audio to Gemini...
Uploaded file: files/abc123xyz
Generating structured prompt...

==================================================
Success! Output saved to: out/1_20251022_165500.md
==================================================

Directory Structure

Voice-To-Prompt-Pipeline/
├── .venv/           # Virtual environment (created by uv)
├── in/              # Place audio files here
├── out/             # Processed prompts output here
├── .env             # Environment variables (API key) - not in git
├── .env.example     # Example environment file
├── .gitignore       # Git ignore rules
├── system-prompt.md # System prompt for Gemini
├── process_audio.py # Main CLI script
├── requirements.txt # Python dependencies
└── README.md        # This file

Security Note

The .env file containing your API key is excluded from git via .gitignore. Never commit your actual API key to version control.