stt-demo

This application demostrates Edge AI's speech-to-text + voice activity detection modules.

Pre-requisites

Install platform dependencies

sudo apt update && sudo apt install -y libasound2 libasound2-data alsa-utils libpython3.11 libopenblas-dev libgomp1 libgfortran5

Models

• edgeai-stt-mini.bin:
  • smallest model
  • 74MB of disk usage
  • 273 MB of memory usage
• edgeai-stt.bin:
  • base model
  • only supports english
  • most efficient
  • 78MB of disk usage
  • 343MN of memory usage
• edgeai-stt-multi.bin
  • support 10+ languages
  • 252MB of disk usage
  • 676MB of memory usage

Usage

Usage: ./stt_app [OPTIONS]

Options:
  -m, --model <path>      Path to the ggml model file (default: ./edgeai-stt.bin)
  -e, --endpoint <url>    HTTP endpoint to post events & output (optional)
  -t, --threads <num>     Number of threads to use (default: 4)
  -g, --gpu               Enable GPU acceleration (Metal/CUDA default: off)
  -h, --help              Show this help message

Example:
  ./stt_app --model ./edgeai-stt.bin --endpoint http://127.0.0.1/agent/run