Files
openwhispr-easy-setup/openwhispr-ptt.py
T
Claude 723436dae9 Switch PTT hotkey from Ctrl+Super to Ctrl+Alt
Super key is intercepted by Cinnamon/window managers before reaching
the app. Ctrl+Alt is a better choice — works reliably on all DEs.

https://claude.ai/code/session_01XKYC1basxdwtHy71tky7xm
2026-03-25 20:01:10 +00:00

122 lines
3.9 KiB
Python

#!/usr/bin/env python3
"""
openwhispr-ptt.py — Push-to-talk wrapper for OpenWhispr
Turns Ctrl+Alt (hold) into Ctrl+` toggle pairs:
- Press Ctrl+Alt → sends Ctrl+` to START dictation
- Release Ctrl+Alt → sends Ctrl+` to STOP dictation
Requires: pynput, xdotool
Usage: python3 openwhispr-ptt.py
"""
import subprocess
import threading
import time
import signal
import sys
try:
from pynput import keyboard
except ImportError:
print("Error: pynput not installed.")
print("Install with: pip3 install pynput")
sys.exit(1)
# ── Configuration ────────────────────────────────────────────────────────────
# The key combo YOU hold to talk
PTT_MODIFIER = keyboard.Key.ctrl_l
PTT_KEY = keyboard.Key.alt_l
# Also accept right-hand variants
PTT_MODIFIER_ALT = keyboard.Key.ctrl_r
PTT_KEY_ALT = keyboard.Key.alt_r
# The key combo OpenWhispr uses (Ctrl+`)
# We simulate this via xdotool for reliability with Electron apps
WHISPR_HOTKEY = "ctrl+grave"
# ── State ────────────────────────────────────────────────────────────────────
is_recording = False
press_time = 0
ctrl_held = False
super_held = False
lock = threading.Lock()
def send_whispr_toggle():
"""Send Ctrl+` to OpenWhispr via xdotool."""
try:
subprocess.run(
["xdotool", "key", WHISPR_HOTKEY],
timeout=2,
capture_output=True,
)
except FileNotFoundError:
print("Error: xdotool not installed. Install with: sudo apt install xdotool")
sys.exit(1)
except subprocess.TimeoutExpired:
pass
def on_press(key):
global is_recording, press_time, ctrl_held, super_held
with lock:
# Track modifier state — ignore key repeat (already True)
if key in (PTT_MODIFIER, PTT_MODIFIER_ALT):
ctrl_held = True
elif key in (PTT_KEY, PTT_KEY_ALT):
super_held = True
# Start recording when both held and not already recording
if ctrl_held and super_held and not is_recording:
press_time = time.time()
is_recording = True
send_whispr_toggle() # START
print("\033[32m● Recording...\033[0m", flush=True)
def on_release(key):
global is_recording, press_time, ctrl_held, super_held
with lock:
released_ptt = False
if key in (PTT_MODIFIER, PTT_MODIFIER_ALT):
ctrl_held = False
released_ptt = True
elif key in (PTT_KEY, PTT_KEY_ALT):
super_held = False
released_ptt = True
# Stop recording when either key is truly released
if is_recording and released_ptt:
elapsed_ms = (time.time() - press_time) * 1000
send_whispr_toggle() # STOP
is_recording = False
print(f"\033[0m○ Stopped ({int(elapsed_ms)}ms)", flush=True)
def main():
print("═══════════════════════════════════════════")
print(" OpenWhispr Push-to-Talk")
print("═══════════════════════════════════════════")
print(" Hold: Ctrl + Alt → start dictation")
print(" Release: Ctrl + Alt → stop dictation")
print(" Quit: Ctrl + C")
print("═══════════════════════════════════════════")
print("Listening...\n")
# Handle Ctrl+C gracefully
signal.signal(signal.SIGINT, lambda *_: sys.exit(0))
with keyboard.Listener(on_press=on_press, on_release=on_release) as listener:
listener.join()
if __name__ == "__main__":
main()