Voice Prompt Enhancement Node
July 30, 2025 ยท View on GitHub
A specialized system prompt for optimizing voice-to-text transcriptions before they're sent to AI inference agents.
Overview
This system prompt provides a model configuration for an AI agent designed to fit into workflows that leverage voice-to-text in automations. The agent acts as a preprocessing layer that transforms raw speech transcriptions into well-structured, optimized prompts.
Example Workflow
Using a service like Voicenotes.com:
- Send transcribed notes tagged as 'prompts' to this processing workflow
- The workflow pipes them through this enhancement agent
- The optimized output is then sent to your main inference agent
Why Use This Approach?
While you could pass raw STT output directly to your AI agent, adding this lightweight processing step significantly improves results:
The Process:
- User sends voice prompt โ STT transcription
- Agent 1 (this enhancement node) recognizes the transcript as a prompt and optimizes it using prompt engineering best practices
- Agent 2 (your inference agent) receives cleaner, more structured input
Result: Better inputs lead to better outputs.
System Prompt
The complete system prompt is available in prompt.md.
Features
- Removes filler words and obvious mistranscriptions
- Improves clarity and coherence
- Organizes content using sections
- Applies prompt engineering best practices
- Creates deterministic, unambiguous prompts