Voice LLM Application - Experiment 03: Multimodal Image Analysis - Voice Dialogue

Experiment Preparation:

  1. Ensure Doubao AI (Volcengine) and iFlytek AI are connected (refer to Experiment 01, Experiment 02)
  2. Find images as experiment material. Image import can be relative or absolute path. Relative path defaults to AI_online_voice/assets/sample.jpg (default relative path image is included in the package, can change relative path image but naming must be sample.jpg)

Experiment Steps: (Ensure voice module is connected)

  1. cd AI_online_voice # Enter main directory
  2. python examples/03_voice_image_dialogue.py # Run example program
  3. After entering the program, according to terminal prompts, first enter y to enter image selection. Can choose absolute or relative path for voice. Absolute path requires manual input, relative path defaults to assets/sample.jpg

Terminal Operation Example:

Relative path selection:

TOOL

Absolute path selection:

TOOL

Image analysis:

TOOL

Experiment Effect: Image upload → Multimodal LLM analysis → Voice dialogue output, implementing image understanding and voice interaction