Audio-to-Data Intelligence Actor

This Apify Actor extracts structured data from audio and video sources (YouTube, Podcasts) using advanced AI models. It can transcribe content and then analyze it to extract specific information like stock tickers, SaaS tool mentions, or key insights.

Features

• Multi-Source Ingestion: Process videos and podcasts from YouTube URLs.
• Flexible Transcription:
  • YouTube Captions (Free): Fast and cost-effective, using existing video captions.
  • OpenAI Whisper (Cloud): High-fidelity transcription (requires OpenAI API Key).
  • Local Whisper (Free): Uses the whisper model on the actor's CPU. Slower, but requires no API key and works if captions are missing.
• AI-Powered Extraction: Use GPT-4o (OpenAI) or Claude 3.5 Sonnet (Anthropic) to structure the data.
• Key-Optional Operation: Can be used without any API keys to simply fetch and output transcripts.
• Customizable Goals: Define exactly what you want to extract (e.g., "stock_tickers", "summary", "names").

Usage

input Configuration

Field	Type	Description	Default
startUrls	Array	List of YouTube or Podcast URLs to process.	[]
extractionGoal	String	Description of the data to extract (e.g., "stock_tickers", "saas_tools").	stock_tickers
transcriptionMode	String	captions (Free) or whisper (Costs $).	captions
aiProvider	String	openai or anthropic.	openai
openaiApiKey	String	Required for Whisper and OpenAI extraction.	null
anthropicApiKey	String	Required for Anthropic extraction.	null

Modes of Operation

1. Transcription Only (Free / No Keys)

If you do not provide an API Key, the Actor will automatically skip the extraction phase.

• Transcription:
  • Tries YouTube Captions first (Fastest).
  • If captions are missing, falls back to Local Whisper (Free, using Actor CPU).
• Output: The raw transcript of the video.
• Cost: Only Apify platform usage.

2. Full Intelligence Extraction

Provide an API Key for your chosen provider (openaiApiKey or anthropicApiKey).

• Output: Structured JSON data based on your extractionGoal.
• Example:

{
  "url": "https://youtube.com/watch?v=...",
  "entityName": "$TSLA",
  "context": "Mentioned as a buy candidate due to AI robotics.",
  "sentiment": "Bullish",
  "aiProvider": "openai"
}

Output

The results are stored in the default Apify dataset. Each item represents an extracted entity or the full transcript if extraction was skipped.

Limits & Pricing

• YouTube Captions: Free.
• Whisper: ~0.006 / minute (via your OpenAI key).
• LLM Extraction: Varies by model usage (GPT-4o / Claude 3.5 Sonnet).

License

Apache 2.0