talk-tailor

TalkTailor

TalkTailor is an open-source, AI-powered web application for transcribing, analyzing, and improving your talks or speeches. It features a modern React frontend and a robust Go backend, leveraging OpenAI's Whisper and GPT models for fast, accurate transcription and advanced text analysis.

Features

• 🎙️ Record or Upload Audio: Record directly in your browser or upload MP3/video files.
• ✨ AI-Powered Transcription: Uses OpenAI Whisper for high-quality, multi-language transcription.
• 📝 Text Correction: Automatic grammar and formatting correction of transcriptions.
• 🧠 Outline & Bulletpoints: Instantly generate speaker outlines and bulletpoints from your transcript using GPT-4.
• 💾 Local Storage: Stores your recent transcriptions and audio securely in your browser.
• 📋 Copy & Edit: Edit, copy, and manage your transcripts with ease.
• 🚀 Modern UI: Built with React and Ant Design for a seamless user experience.
• 🐳 Docker Support: Easy deployment with Docker.

Architecture

graph TD
  A[User: Browser] -->|Audio Upload/Record| B(React Frontend)
  B -->|REST API| C(Go Backend)
  C -->|Transcription| D(OpenAI Whisper API)
  C -->|Outline/Bulletpoints| E(OpenAI GPT API)
  B -->|Transcription, Outline, Bulletpoints| A

Demo

1. Record or Upload: Start a new recording or upload an audio/video file.
2. Transcribe: Let the AI transcribe and correct your speech.
3. Analyze: Generate outlines or bulletpoints for your talk.
4. Edit & Copy: Refine and copy your transcript for further use.

Getting Started

Prerequisites

• Go 1.20+
• Node.js 18+
• npm or yarn
• Docker (optional, for containerized deployment)
• OpenAI API Key (get one here)

Clone the Repository

git clone https://github.com/icereed/talk-tailor.git
cd talk-tailor

Local Development

1. Backend (Go)

Set your OpenAI API key:

export OPENAI_API_KEY=your-openai-key

Install dependencies and run the server:

go mod tidy
go run main.go

The backend will serve the frontend and expose API endpoints at http://localhost:8080.

2. Frontend (React)

cd client
npm install
npm run build

The production build will be placed in client/dist and served by the Go backend.

Docker Usage

You can run the latest image directly from GitHub Container Registry (GHCR):

docker run -e OPENAI_API_KEY=your-openai-key -p 8080:8080 ghcr.io/icereed/talk-tailor:latest

Or build and run locally:

docker build -t talk-tailor .
docker run -e OPENAI_API_KEY=your-openai-key -p 8080:8080 talk-tailor

Image source: ghcr.io/icereed/talk-tailor

API Reference

POST /api/transcribe

• Description: Upload an audio file (MP3 or video) to receive a transcription.
• Request: multipart/form-data with audio file field.
• Response: JSON with original and corrected transcription.

POST /api/outline

• Description: Generate a detailed speaker outline from transcript text.
• Request: JSON { "text": "..." }
• Response: JSON { "response": "..." }

POST /api/bulletpoints

• Description: Convert transcript text into bulletpoints.
• Request: JSON { "text": "..." }
• Response: JSON { "response": "..." }

Configuration

• OPENAI_API_KEY (required): Your OpenAI API key for transcription and text analysis.

Contributing

Contributions are welcome! Please open issues or pull requests for bug fixes, features, or improvements.

• Follow standard Go and React/TypeScript best practices.
• Ensure code is well-documented and tested.
• By contributing, you agree to license your work under the MIT License.

Roadmap

The following roadmap outlines possible directions for community-driven development:

1. Customizability

• Custom AI Actions: Define your own AI-powered actions (e.g., "Summarize for LinkedIn", "Extract Q&A", "Generate Quiz") via prompt templates.
• Personalized Feedback: Set personal goals (e.g., pacing, filler word reduction) and receive targeted feedback.
• UI Themes & Layouts: Dark mode, font size, and customizable dashboard widgets.
• Language & Region Settings: Multi-language UI and locale-aware formatting.

2. Advanced AI Enhancements

• Speech Analytics: Detect filler words, pacing, pauses, and provide visual analytics. Sentiment/emotion analysis of speech.
• Speaker Coaching: AI-generated tips based on transcript analysis (clarity, engagement). Simulated audience Q&A and answer quality rating.
• Custom Model Integration: Plug in your own OpenAI-compatible models or endpoints. Support for fine-tuned models for specific domains.

3. Training & Practice Tools

• Practice Mode: Timed speaking exercises with real-time feedback.
• Progress Tracking: Visualize improvement over time (charts, badges, milestones).
• Peer Review: Share recordings/transcripts for community or mentor feedback.
• Scenario Library: Pre-built scenarios (e.g., job interview, TED talk) with tailored feedback.

4. Collaboration & Sharing

• Team Workspaces: Shared libraries for teams, clubs, or classes.
• Export Options: Export to PDF, DOCX, slides, or share via link.
• API & Plugin System: Third-party integrations (calendar, LMS, video platforms).

5. Privacy & Data Control

• On-Premise/Private Mode: Option to run all processing locally or on self-hosted infrastructure.
• Granular Data Controls: User control over data retention, sharing, and deletion.

License

This project is licensed under the MIT License.

Acknowledgements

• OpenAI Whisper
• OpenAI GPT
• Gin Web Framework
• Ant Design