rnentjes/speech2text
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Add Alt key support for toggling transcription language dynamically

Browse Source
This commit is contained in:
rnentjes
2026-02-16 14:18:24 +01:00
parent fb51972ec9
commit 4706a4a8a6
1 changed files with 27 additions and 5 deletions
Download Patch File Download Diff File Expand all files Collapse all files

32
speech2text-keyed.py
View File

@@ -76,6 +76,9 @@ class SpeechApp(Gtk.Window):
self.audio = []
self.hotkey_pressed = set()
self.hotkey_active = False
self.lang_modifier_active = False
self.default_language = "en"
self.alt_language = "nl"
# Global hotkey listener for Right-Shift + Right-Ctrl
self.listener = keyboard.Listener(
@@ -115,7 +118,8 @@ class SpeechApp(Gtk.Window):
self.record_thread.join()
# Update status bar
self.update_status("Transcribing...")
threading.Thread(target=self.transcribe_and_type).start()
language = self.alt_language if self.lang_modifier_active else self.default_language
threading.Thread(target=self.transcribe_and_type, args=(language,)).start()
def record_audio(self):
def callback(indata, frames, time_info, status):
@@ -141,7 +145,7 @@ class SpeechApp(Gtk.Window):
self.status_bar.set_text(f"Status: {status}")
return False # Return False to prevent being called again
def transcribe_and_type(self):
def transcribe_and_type(self, language):
if not self.audio:
# Update status back to Ready if no audio was recorded
GLib.idle_add(self.update_status, "Ready")
@@ -157,7 +161,7 @@ class SpeechApp(Gtk.Window):
# In the transcribe_and_type method, modify the transcribe call:
# Replace "en" with your desired language code
result = model.transcribe(tmpfile.name, language="en")
result = model.transcribe(tmpfile.name, language=language)
text = result["text"].strip()
# Display the text in the text area
@@ -208,8 +212,14 @@ class SpeechApp(Gtk.Window):
self.hotkey_pressed.add("shift_r")
elif key == keyboard.Key.ctrl_r:
self.hotkey_pressed.add("ctrl_r")
elif key == keyboard.Key.alt_r:
self.hotkey_pressed.add("alt_r")
if not self.hotkey_active and self.hotkey_pressed == {"shift_r", "ctrl_r"}:
self.lang_modifier_active = (
"alt_r" in self.hotkey_pressed
)
if not self.hotkey_active and (self.hotkey_pressed == {"shift_r", "ctrl_r"} or self.hotkey_pressed == {"shift_r", "alt_r"}):
self.hotkey_active = True
GLib.idle_add(self.start_recording)
@@ -218,11 +228,23 @@ class SpeechApp(Gtk.Window):
self.hotkey_pressed.discard("shift_r")
elif key == keyboard.Key.ctrl_r:
self.hotkey_pressed.discard("ctrl_r")
elif key == keyboard.Key.alt_r:
self.hotkey_pressed.discard("alt_r")
if self.hotkey_active and self.hotkey_pressed != {"shift_r", "ctrl_r"}:
self.lang_modifier_active = (
"alt_r" in self.hotkey_pressed
)
if self.hotkey_active and (self.hotkey_pressed != {"shift_r", "ctrl_r"} and self.hotkey_pressed != {"shift_r", "alt_r"}):
self.hotkey_active = False
GLib.idle_add(self.stop_recording)
def is_shift_key(self, key):
return key in (keyboard.Key.shift, keyboard.Key.shift_l, keyboard.Key.shift_r)
def is_alt_key(self, key):
return key in (keyboard.Key.alt, keyboard.Key.alt_l, keyboard.Key.alt_r, keyboard.Key.alt_gr)
def on_destroy(self, widget):
try:
self.listener.stop()