Manifest

ability-voice

Voice operations - TTS (Piper), STT (Whisper), wake word detection, optimized for NVIDIA Jetson

Field	Value
Version	1.0.0
Type	ability
Entrypoint	src/ability_voice/__main__.py

Abilities

• secret-ability ^0.4.1

Brokers

• remote: wss://broker.dadavidtseng.com/kadi

Scripts

kadi run preflight  # python3 -c "import torch; print(f'CUDA available: {torch.cuda.is_available()}')" || python3 --version
kadi run setup  # python3 -m venv venv && ./venv/bin/pip install --upgrade pip && ./venv/bin/pip install -r requirements.txt
kadi run start  # ./venv/bin/python3 -m ability_voice
kadi run listen  # LISTEN_MODE=1 ./venv/bin/python3 -m ability_voice
kadi run standalone  # STANDALONE=1 ./venv/bin/python3 -m ability_voice
kadi run test  # ./venv/bin/python3 -m pytest tests/
kadi run clean  # rm -rf venv .venv __pycache__ src/ability_voice/__pycache__

Deploy Profiles

local

• Target: local
• Engine: docker
• Services: voice

local-listen

• Target: local
• Engine: docker
• Services: voice