Add push-to-talk wrapper for OpenWhispr

Python script that converts Ctrl+Super hold-to-talk into Ctrl+`
toggle pairs for OpenWhispr, working around the app's broken
Hold activation mode on Linux.

Uses pynput for key listening and xdotool to send hotkeys.

https://claude.ai/code/session_01XKYC1basxdwtHy71tky7xm
This commit is contained in:
Claude
2026-03-25 19:56:43 +00:00
parent cc847cff67
commit e32dc5f690
+114
View File
@@ -0,0 +1,114 @@
#!/usr/bin/env python3
"""
openwhispr-ptt.py — Push-to-talk wrapper for OpenWhispr
Turns Ctrl+Super (hold) into Ctrl+` toggle pairs:
- Press Ctrl+Super → sends Ctrl+` to START dictation
- Release Ctrl+Super → 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.cmd # Super/Windows key
# The key combo OpenWhispr uses (Ctrl+`)
# We simulate this via xdotool for reliability with Electron apps
WHISPR_HOTKEY = "ctrl+grave"
# Minimum hold time (ms) to avoid accidental triggers
MIN_HOLD_MS = 100
# ── State ────────────────────────────────────────────────────────────────────
pressed_keys = set()
is_recording = False
press_time = 0
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
with lock:
pressed_keys.add(key)
# Check if both PTT keys are held
if PTT_MODIFIER in pressed_keys and PTT_KEY in pressed_keys:
if not is_recording:
press_time = time.time()
is_recording = True
send_whispr_toggle() # START
print("\033[32m● Recording...\033[0m")
def on_release(key):
global is_recording, press_time
with lock:
# If either PTT key is released while recording
if is_recording and key in (PTT_MODIFIER, PTT_KEY):
elapsed_ms = (time.time() - press_time) * 1000
if elapsed_ms >= MIN_HOLD_MS:
send_whispr_toggle() # STOP
print(f"\033[0m○ Stopped ({int(elapsed_ms)}ms)")
else:
# Too short — send another toggle to cancel
send_whispr_toggle()
print(f"\033[33m○ Cancelled (too short: {int(elapsed_ms)}ms)\033[0m")
is_recording = False
pressed_keys.discard(key)
def main():
print("═══════════════════════════════════════════")
print(" OpenWhispr Push-to-Talk")
print("═══════════════════════════════════════════")
print(f" Hold: Ctrl + Super → start dictation")
print(f" Release: Ctrl + Super → stop dictation")
print(f" 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()