Scenario Configuration

What is Scenario-based Configuration

A scenario is a set of preset LLM prompts and processing rules. In different scenarios, the same voice content will be processed into different style results.

For example:

• Meeting Notes → Structured minutes with key topics and action items
• Email Writing → Formal email format with proper paragraphs
• Simple Polish → Basic cleanup—typos, punctuation, formatting

Built-in Scenarios

1. Meeting Notes

Organize spoken content into meeting minutes, automatically extracting key topics and action items.

2. Email Writing

Convert conversational input to formal email format, automatically organizing paragraphs.

3. Simple Polish

Simple polishing of transcription results—typos, incorrect expressions, punctuation—keeping the original conversational feel.

4. Professional Polish

More formal processing, re-expressing based on original meaning to get logically clear results.

Custom Scenarios

In scenario settings, click "New Scenario" and enter:

• Scenario name
• Custom prompt

The system provides default prompts, but you can fully customize them.

LLM Configuration

Voconly supports LLM post-processing with:

Local Solutions

• Llama.cpp: Directly load local GGUF models
• Ollama: Run local models through Ollama

Cloud APIs

• OpenAI
• DeepSeek
• Qwen
• Claude
• Zhipu AI

Just enter the corresponding API key in settings to enable.

No LLM

You can also use Voconly for pure speech-to-text without LLM processing.

Common Issues

• Low recognition rate: Check microphone, try a larger model
• Slow transcription: Ensure GPU drivers are installed, or use a smaller model
• Slow LLM processing: Local models need more VRAM, or switch to cloud API