🤖 AI Assistant Integration

Process your recognition results using models deployed in Ollama. Customize templates to handle various tasks.

Last Updated: 2026-03-31 · Language: English

1. Download & Install Ollama

Visit https://ollama.com/download

1. Install Ollama on your local machine and start it.
2. Verify the service address in AI Settings (default is http://localhost:11434/api).
3. Click "Test" and ensure the status is "Available" before proceeding.
4. Search for models at https://ollama.com/search and pull the model you need (e.g., qwen2.5:1.5b).

2. Task Types

• Translation: Translate text into other languages.
• Correction: Fix typos and grammatical errors in the transcript.
• Summary: Generate meeting minutes or a brief summary.
• Custom: Define your own prompts to let the LLM process results according to your specific needs.

3. Input Modes

• Single: Process segments one by one (this is the only mode available when calling LLM during real-time transcription).
• Batch: Process data in batches.
• All: Process the entire transcript at once, ideal for full-text summaries.

4. Custom Model Parameters

• Model parameters affect the LLM output. Professionals can fine-tune these settings as needed.
• Built-in support for common parameters; toggle "Model Parameters" to customize.

5. FAQ

• Cannot connect to model: Check if Ollama is running and the address is correct.
• Slow output: Try a model with fewer parameters (e.g., 1.5B or 4B models).

6. Recommendations

• For Translation: We recommend the HY-MT1.5-1.8B model. It's a professional translation model supporting multiple languages with fast inference speeds.