bossiara13/J.A.R.V.I.S-py
1
1
Fork 0
Code Issues Pull requests Projects Releases 6 Packages Activity Actions

feat: add webrtcvad-based smart command-end detection

Browse source 
Replace the fixed 10s post-wake-word listen window with a VAD-driven
loop. Once at least 500ms of speech have been seen, 1.2s of silence
closes the window. Hard cap at 15s as a safety net.

Tunable in config.py via VAD_AGGRESSIVENESS, COMMAND_END_SILENCE_MS,
COMMAND_MIN_SPEECH_MS, COMMAND_MIN_LISTEN_MS, COMMAND_MAX_LISTEN_MS.
This commit is contained in:
Bossiara13 2026-04-22 19:32:30 +03:00
parent 277921ce14
commit 794d923bf9
3 changed files with 45 additions and 6 deletions
Download patch file Download diff file Expand all files Collapse all files

8
config.py
View file

@ -26,3 +26,11 @@ CHROME_PATH = 'C:/Program Files (x86)/Google/Chrome/Application/chrome.exe %s'
GROQ_TOKEN = os.getenv('GROQ_TOKEN')
GROQ_BASE_URL = "https://api.groq.com/openai/v1"
GROQ_MODEL = "llama-3.3-70b-versatile"
# VAD (webrtcvad) — определяет конец команды по тишине вместо фиксированных 10 сек.
# 0..3, выше — агрессивнее режет шум (но и обрезает речь).
VAD_AGGRESSIVENESS = 2
COMMAND_END_SILENCE_MS = 1200
COMMAND_MIN_SPEECH_MS = 500
COMMAND_MIN_LISTEN_MS = 1000
COMMAND_MAX_LISTEN_MS = 15000