This project provides a "Podcast Copilot" for generating social media posts and accompanying images to promote podcast episodes. It targets podcast creators and social media managers, automating content creation by leveraging a chain of AI models.

How It Works

The copilot orchestrates a series of AI models via LangChain. It begins with Whisper for speech-to-text transcription, then uses Dolly 2 to identify guest names. Bing Search grounds guest biographies, and GPT-4 generates promotional social media copy and a DALL-E prompt. DALL-E creates an image from the prompt, and a LinkedIn plugin can post the content. This modular, chained approach allows for flexible integration of different AI capabilities.

Quick Start & Requirements

• Install dependencies: pip install -r requirements.txt
• Run script: python PodcastSocialMediaCopilot.py
• Requires an Azure OpenAI resource with deployments for gpt-4, dalle, and a plugins-capable model.
• Requires a Bing Search resource.
• Configuration involves updating API keys and resource names in PodcastSocialMediaCopilot.py.
• Access to DALL-E models and plugins-capable models may require requesting access via Azure.

Highlighted Details

• Demonstrated at Microsoft Build 2023 by CTO Kevin Scott.
• Integrates speech-to-text, guest extraction, web grounding, text generation, and image generation.
• Utilizes LangChain for model orchestration.
• Includes a LinkedIn posting plugin.

Maintenance & Community

• Contributions are welcome, subject to a Contributor License Agreement (CLA).
• Follows the Microsoft Open Source Code of Conduct.

Licensing & Compatibility

• The README does not explicitly state a license.
• Compatibility for commercial use or closed-source linking is not specified.

Limitations & Caveats

• DALL-E and plugins-capable models were in private preview at the time of the demo (May 2023) and may require specific access requests.
• The setup is heavily reliant on Azure cloud services and specific model deployments.