tomatocream/tooling
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

test

Browse Source
This commit is contained in:
dingfeng.wong
2025-07-22 22:11:37 +08:00
parent af840601c0
commit 0ad663c835
3 changed files with 418 additions and 285 deletions
Download Patch File Download Diff File Expand all files Collapse all files
STT_STATUSBAR_USAGE.md
+116
View File
@@ -0,0 +1,116 @@
# STT Status Bar App Usage
## Overview
The STT (Speech-to-Text) status bar app provides a convenient macOS menu bar interface for controlling speech-to-text functionality using wake word activation.
## Features
- **Start/Stop/Pause/Resume**: Full control over STT recording
- **Status Indicators**: Visual status in menu bar (🎙️🔴 recording, 🎙️⏸️ paused, 🎙️⚫ stopped)
- **Configurable Settings**: Change wake words and models on the fly
- **File Output**: Save transcriptions to a file
- **Notifications**: Real-time notifications for transcriptions and status changes
## Usage
### Starting the Status Bar App
```bash
# Launch the status bar app
tooling stt statusbar
# Or using the full CLI path
python -m tooling.cli stt statusbar
```
### Menu Options
#### Main Controls
- **Start STT**: Begin speech-to-text with current settings
- **Stop STT**: Stop speech-to-text completely
- **Pause**: Temporarily pause recognition (keeps recorder alive)
- **Resume**: Resume recognition from pause
#### Settings
- **Wake Word**: Choose from predefined options:
- jarvis (default)
- alexa
- hey google
- hey siri
- computer
- **Model**: Select Whisper model:
- tiny (fastest, least accurate)
- base (default, good balance)
- small (more accurate)
- medium (most accurate, slower)
#### File Management
- **Show Recent Transcriptions**: Display last 10 transcription entries
- **Save to File...**: Set output file for saving transcriptions
### Status Indicators
| Icon | Status | Description |
|------|--------|-------------|
| 🎙️⚫ | Stopped | STT is not running |
| 🎙️🔴 | Recording | STT is active and listening |
| 🎙️⏸️ | Paused | STT is paused but can be resumed |
### Workflow Example
1. **Launch the app**: `tooling stt statusbar`
2. **Look for the 🎙️ icon** in your macOS menu bar
3. **Click the icon** to open the menu
4. **Set your preferences**:
- Choose wake word from Settings > Wake Word
- Select model from Settings > Model
- Optionally set output file with "Save to File..."
5. **Click "Start STT"** to begin
6. **Say your wake word** (e.g., "jarvis") to trigger recording
7. **Speak clearly** after the wake word is detected
8. **Get notifications** with your transcribed text
9. **Use Pause/Resume** as needed
10. **Click "Stop STT"** when done
### Notifications
The app provides notifications for:
- STT started/stopped/paused/resumed
- Transcribed speech (shows first 100 characters)
- Settings changes (when STT is running)
- File operations
### File Output
When you set an output file:
- Transcriptions are saved with timestamps
- Sessions are marked with start/end timestamps
- File is automatically created in ~/Documents/ by default
- Each transcription includes: `[HH:MM:SS] transcribed text`
### Requirements
- macOS (rumps requires macOS and PyObjC)
- RealtimeSTT library
- Working microphone
- Python 3.11+
### Troubleshooting
- **No menu bar icon**: Make sure you're running on macOS and rumps is installed
- **No transcriptions**: Check microphone permissions and try speaking louder
- **Wake word not detected**: Try adjusting sensitivity or use a different wake word
- **High CPU usage**: Consider using the "tiny" model for better performance
### Advanced Configuration
The status bar app uses sensible defaults, but you can modify the underlying configuration by editing the STTStatusBarApp class in `src/tooling/stt_cli.py`.
Default settings:
- Wake word: "jarvis"
- Model: "base"
- Sensitivity: 0.6
- Device: auto-detect (CUDA if available, otherwise CPU)
- Realtime display: enabled
realtimesst.log
+17
View File
@@ -0,0 +1,17 @@
2025-07-22 22:10:54.507 - RealTimeSTT: realtimestt - INFO - Starting RealTimeSTT
2025-07-22 22:10:54.520 - RealTimeSTT: realtimestt - INFO - Initializing audio recording (creating pyAudio input stream, sample rate: 16000 buffer size: 512
2025-07-22 22:10:54.523 - RealTimeSTT: realtimestt - INFO - Initializing faster_whisper realtime transcription model tiny, default device: cpu, compute type: default, device index: 0, download root: None
2025-07-22 22:10:55.181 - RealTimeSTT: realtimestt - DEBUG - Faster_whisper realtime speech to text transcription model initialized successfully
2025-07-22 22:10:55.181 - RealTimeSTT: realtimestt - ERROR - Wakeword engine unknown/unsupported or wake_words not specified. Please specify one of: pvporcupine, openwakeword.
NoneType: None
2025-07-22 22:10:55.181 - RealTimeSTT: realtimestt - INFO - Initializing WebRTC voice with Sensitivity 3
2025-07-22 22:10:55.181 - RealTimeSTT: realtimestt - DEBUG - WebRTC VAD voice activity detection engine initialized successfully
2025-07-22 22:10:55.838 - RealTimeSTT: realtimestt - DEBUG - Silero VAD voice activity detection engine initialized successfully
2025-07-22 22:10:55.838 - RealTimeSTT: realtimestt - DEBUG - Starting realtime worker
2025-07-22 22:10:55.838 - RealTimeSTT: realtimestt - DEBUG - Waiting for main transcription model to start
2025-07-22 22:11:01.946 - RealTimeSTT: realtimestt - DEBUG - Main transcription model ready
2025-07-22 22:11:01.946 - RealTimeSTT: realtimestt - DEBUG - RealtimeSTT initialization completed successfully
2025-07-22 22:11:01.946 - RealTimeSTT: realtimestt - INFO - Setting listen time
2025-07-22 22:11:01.946 - RealTimeSTT: realtimestt - INFO - State changed from 'inactive' to 'listening'
2025-07-22 22:11:01.947 - RealTimeSTT: realtimestt - DEBUG - Waiting for recording start
2025-07-22 22:11:01.981 - RealTimeSTT: realtimestt - INFO - State changed from 'listening' to 'wakeword'
src/tooling/stt_cli.py
+285 -285
View File
@@ -96,300 +96,300 @@ if RUMPS_AVAILABLE:
def __init__(self, name="STT", **kwargs):
super().__init__(name, **kwargs)
# Initialize state
self.recorder = None
self.is_running = False
self.is_paused = False
self.output_file = None
self.transcription_thread = None
self.stop_event = threading.Event()
# Configuration
self.wake_word = "jarvis"
self.model = "base"
self.language = ""
self.realtime = True
self.sensitivity = 0.6
self.device = "auto"
self.save_to_file = None
# Menu setup
self.menu = [
rumps.MenuItem("Start STT", callback=self.start_stt),
rumps.MenuItem("Stop STT", callback=self.stop_stt),
rumps.separator,
rumps.MenuItem("Pause", callback=self.pause_stt),
rumps.MenuItem("Resume", callback=self.resume_stt),
rumps.separator,
{
"Settings": {
"Wake Word": {
"jarvis": rumps.MenuItem("jarvis", callback=self.set_wake_word),
"alexa": rumps.MenuItem("alexa", callback=self.set_wake_word),
"hey google": rumps.MenuItem("hey google", callback=self.set_wake_word),
"hey siri": rumps.MenuItem("hey siri", callback=self.set_wake_word),
"computer": rumps.MenuItem("computer", callback=self.set_wake_word),
},
"Model": {
"tiny": rumps.MenuItem("tiny", callback=self.set_model),
"base": rumps.MenuItem("base", callback=self.set_model),
"small": rumps.MenuItem("small", callback=self.set_model),
"medium": rumps.MenuItem("medium", callback=self.set_model),
}
}
},
rumps.separator,
rumps.MenuItem("Show Recent Transcriptions", callback=self.show_transcriptions),
rumps.MenuItem("Save to File...", callback=self.select_output_file),
rumps.separator,
rumps.MenuItem("Quit", callback=rumps.quit_application),
]
# Set initial menu states
self.update_menu_states()
self.set_current_wake_word()
self.set_current_model()
def update_menu_states(self):
"""Update menu item states based on current status."""
if self.is_running:
if self.is_paused:
self.title = "🎙️⏸️"
self.menu["Start STT"].set_callback(None)
self.menu["Stop STT"].set_callback(self.stop_stt)
self.menu["Pause"].set_callback(None)
self.menu["Resume"].set_callback(self.resume_stt)
else:
self.title = "🎙️🔴"
self.menu["Start STT"].set_callback(None)
self.menu["Stop STT"].set_callback(self.stop_stt)
self.menu["Pause"].set_callback(self.pause_stt)
self.menu["Resume"].set_callback(None)
else:
self.title = "🎙️⚫"
self.menu["Start STT"].set_callback(self.start_stt)
self.menu["Stop STT"].set_callback(None)
self.menu["Pause"].set_callback(None)
self.menu["Resume"].set_callback(None)
def set_current_wake_word(self):
"""Set checkmark for current wake word."""
for word in ["jarvis", "alexa", "hey google", "hey siri", "computer"]:
self.menu["Settings"]["Wake Word"][word].state = (word == self.wake_word)
def set_current_model(self):
"""Set checkmark for current model."""
for model in ["tiny", "base", "small", "medium"]:
self.menu["Settings"]["Model"][model].state = (model == self.model)
def set_wake_word(self, sender):
"""Set the wake word from menu selection."""
self.wake_word = sender.title
self.set_current_wake_word()
if self.is_running:
rumps.notification("STT Settings", "Wake Word Changed", f"Restart STT to use '{self.wake_word}'")
def set_model(self, sender):
"""Set the model from menu selection."""
self.model = sender.title
self.set_current_model()
if self.is_running:
rumps.notification("STT Settings", "Model Changed", f"Restart STT to use '{self.model}' model")
def start_stt(self, _):
"""Start STT functionality."""
if self.is_running:
return
try:
from RealtimeSTT import AudioToTextRecorder
except ImportError:
rumps.alert("RealtimeSTT not installed", "Please install with: pip install RealtimeSTT")
return
try:
# Determine device
if self.device == "auto":
try:
import torch
self.device = "cuda" if torch.cuda.is_available() else "cpu"
except ImportError:
self.device = "cpu"
# Setup output file if specified
if self.save_to_file:
self.save_to_file.parent.mkdir(parents=True, exist_ok=True)
self.output_file = open(self.save_to_file, 'a', encoding='utf-8')
timestamp = datetime.datetime.now().strftime("%Y-%m-%d %H:%M:%S")
self.output_file.write(f"\n=== STT Session Started: {timestamp} ===\n")
self.output_file.flush()
# Configure recorder
recorder_config = {
"model": self.model,
"wake_words": self.wake_word,
"wake_words_sensitivity": self.sensitivity,
"device": self.device,
"on_recording_start": self.on_recording_start,
"on_recording_stop": self.on_recording_stop,
"on_wakeword_detected": self.on_wakeword_detected,
"on_wakeword_timeout": self.on_wakeword_timeout,
"on_wakeword_detection_start": self.on_wakeword_detection_start,
}
if self.language:
recorder_config["language"] = self.language
if self.realtime:
recorder_config.update({
"enable_realtime_transcription": True,
"on_realtime_transcription_update": self.on_realtime_transcription,
})
# Initialize recorder
self.recorder = AudioToTextRecorder(**recorder_config)
# Start transcription thread
self.is_running = True
# Initialize state
self.recorder = None
self.is_running = False
self.is_paused = False
self.stop_event.clear()
self.transcription_thread = threading.Thread(target=self.transcription_loop, daemon=True)
self.transcription_thread.start()
self.output_file = None
self.transcription_thread = None
self.stop_event = threading.Event()
# Configuration
self.wake_word = "jarvis"
self.model = "base"
self.language = ""
self.realtime = True
self.sensitivity = 0.6
self.device = "auto"
self.save_to_file = None
# Menu setup
self.menu = [
rumps.MenuItem("Start STT", callback=self.start_stt),
rumps.MenuItem("Stop STT", callback=self.stop_stt),
rumps.separator,
rumps.MenuItem("Pause", callback=self.pause_stt),
rumps.MenuItem("Resume", callback=self.resume_stt),
rumps.separator,
{
"Settings": {
"Wake Word": {
"jarvis": rumps.MenuItem("jarvis", callback=self.set_wake_word),
"alexa": rumps.MenuItem("alexa", callback=self.set_wake_word),
"hey google": rumps.MenuItem("hey google", callback=self.set_wake_word),
"hey siri": rumps.MenuItem("hey siri", callback=self.set_wake_word),
"computer": rumps.MenuItem("computer", callback=self.set_wake_word),
},
"Model": {
"tiny": rumps.MenuItem("tiny", callback=self.set_model),
"base": rumps.MenuItem("base", callback=self.set_model),
"small": rumps.MenuItem("small", callback=self.set_model),
"medium": rumps.MenuItem("medium", callback=self.set_model),
}
}
},
rumps.separator,
rumps.MenuItem("Show Recent Transcriptions", callback=self.show_transcriptions),
rumps.MenuItem("Save to File...", callback=self.select_output_file),
rumps.separator,
rumps.MenuItem("Quit", callback=rumps.quit_application),
]
# Set initial menu states
self.update_menu_states()
self.set_current_wake_word()
self.set_current_model()
def update_menu_states(self):
"""Update menu item states based on current status."""
if self.is_running:
if self.is_paused:
self.title = "🎙️⏸️"
self.menu["Start STT"].set_callback(None)
self.menu["Stop STT"].set_callback(self.stop_stt)
self.menu["Pause"].set_callback(None)
self.menu["Resume"].set_callback(self.resume_stt)
else:
self.title = "🎙️🔴"
self.menu["Start STT"].set_callback(None)
self.menu["Stop STT"].set_callback(self.stop_stt)
self.menu["Pause"].set_callback(self.pause_stt)
self.menu["Resume"].set_callback(None)
else:
self.title = "🎙️⚫"
self.menu["Start STT"].set_callback(self.start_stt)
self.menu["Stop STT"].set_callback(None)
self.menu["Pause"].set_callback(None)
self.menu["Resume"].set_callback(None)
def set_current_wake_word(self):
"""Set checkmark for current wake word."""
for word in ["jarvis", "alexa", "hey google", "hey siri", "computer"]:
self.menu["Settings"]["Wake Word"][word].state = (word == self.wake_word)
def set_current_model(self):
"""Set checkmark for current model."""
for model in ["tiny", "base", "small", "medium"]:
self.menu["Settings"]["Model"][model].state = (model == self.model)
def set_wake_word(self, sender):
"""Set the wake word from menu selection."""
self.wake_word = sender.title
self.set_current_wake_word()
if self.is_running:
rumps.notification("STT Settings", "Wake Word Changed", f"Restart STT to use '{self.wake_word}'")
def set_model(self, sender):
"""Set the model from menu selection."""
self.model = sender.title
self.set_current_model()
if self.is_running:
rumps.notification("STT Settings", "Model Changed", f"Restart STT to use '{self.model}' model")
def start_stt(self, _):
"""Start STT functionality."""
if self.is_running:
return
try:
from RealtimeSTT import AudioToTextRecorder
except ImportError:
rumps.alert("RealtimeSTT not installed", "Please install with: pip install RealtimeSTT")
return
try:
# Determine device
if self.device == "auto":
try:
import torch
self.device = "cuda" if torch.cuda.is_available() else "cpu"
except ImportError:
self.device = "cpu"
# Setup output file if specified
if self.save_to_file:
self.save_to_file.parent.mkdir(parents=True, exist_ok=True)
self.output_file = open(self.save_to_file, 'a', encoding='utf-8')
timestamp = datetime.datetime.now().strftime("%Y-%m-%d %H:%M:%S")
self.output_file.write(f"\n=== STT Session Started: {timestamp} ===\n")
self.output_file.flush()
# Configure recorder
recorder_config = {
"model": self.model,
"wake_words": self.wake_word,
"wake_words_sensitivity": self.sensitivity,
"device": self.device,
"on_recording_start": self.on_recording_start,
"on_recording_stop": self.on_recording_stop,
"on_wakeword_detected": self.on_wakeword_detected,
"on_wakeword_timeout": self.on_wakeword_timeout,
"on_wakeword_detection_start": self.on_wakeword_detection_start,
}
if self.language:
recorder_config["language"] = self.language
if self.realtime:
recorder_config.update({
"enable_realtime_transcription": True,
"on_realtime_transcription_update": self.on_realtime_transcription,
})
# Initialize recorder
self.recorder = AudioToTextRecorder(**recorder_config)
# Start transcription thread
self.is_running = True
self.is_paused = False
self.stop_event.clear()
self.transcription_thread = threading.Thread(target=self.transcription_loop, daemon=True)
self.transcription_thread.start()
self.update_menu_states()
rumps.notification("STT Started", f"Wake word: {self.wake_word}", f"Model: {self.model} | Device: {self.device}")
except Exception as e:
rumps.alert("STT Error", f"Failed to start STT: {str(e)}")
def stop_stt(self, _):
"""Stop STT functionality."""
if not self.is_running:
return
self.is_running = False
self.is_paused = False
self.stop_event.set()
if self.recorder:
try:
self.recorder.shutdown()
except:
pass
self.recorder = None
if self.output_file:
timestamp = datetime.datetime.now().strftime("%Y-%m-%d %H:%M:%S")
self.output_file.write(f"=== STT Session Ended: {timestamp} ===\n\n")
self.output_file.close()
self.output_file = None
self.update_menu_states()
rumps.notification("STT Started", f"Wake word: {self.wake_word}", f"Model: {self.model} | Device: {self.device}")
rumps.notification("STT Stopped", "Speech-to-text has been stopped", "")
def pause_stt(self, _):
"""Pause STT functionality."""
if not self.is_running or self.is_paused:
return
except Exception as e:
rumps.alert("STT Error", f"Failed to start STT: {str(e)}")
def stop_stt(self, _):
"""Stop STT functionality."""
if not self.is_running:
return
self.is_paused = True
self.update_menu_states()
rumps.notification("STT Paused", "Speech recognition paused", "Resume from menu")
self.is_running = False
self.is_paused = False
self.stop_event.set()
if self.recorder:
try:
self.recorder.shutdown()
except:
pass
self.recorder = None
if self.output_file:
timestamp = datetime.datetime.now().strftime("%Y-%m-%d %H:%M:%S")
self.output_file.write(f"=== STT Session Ended: {timestamp} ===\n\n")
self.output_file.close()
self.output_file = None
self.update_menu_states()
rumps.notification("STT Stopped", "Speech-to-text has been stopped", "")
def pause_stt(self, _):
"""Pause STT functionality."""
if not self.is_running or self.is_paused:
return
self.is_paused = True
self.update_menu_states()
rumps.notification("STT Paused", "Speech recognition paused", "Resume from menu")
def resume_stt(self, _):
"""Resume STT functionality."""
if not self.is_running or not self.is_paused:
return
self.is_paused = False
self.update_menu_states()
rumps.notification("STT Resumed", "Speech recognition resumed", f"Listening for '{self.wake_word}'")
def transcription_loop(self):
"""Main transcription loop running in background thread."""
while self.is_running and not self.stop_event.is_set():
try:
if not self.is_paused and self.recorder:
text = self.recorder.text()
if text and text.strip():
self.on_transcription_complete(text)
else:
time.sleep(0.1) # Small delay when paused
except Exception as e:
print(f"Transcription error: {e}")
time.sleep(1) # Longer delay on error
def on_realtime_transcription(self, text: str):
"""Handle real-time transcription updates."""
# Could update a notification or log
pass
def on_transcription_complete(self, text: str):
"""Handle completed transcriptions."""
if text.strip():
# Show notification with transcription
rumps.notification("Transcription", "Speech detected:", text[:100] + ("..." if len(text) > 100 else ""))
def resume_stt(self, _):
"""Resume STT functionality."""
if not self.is_running or not self.is_paused:
return
# Save to file if specified
if self.output_file:
timestamp = datetime.datetime.now().strftime("%H:%M:%S")
self.output_file.write(f"[{timestamp}] {text}\n")
self.output_file.flush()
def on_recording_start(self):
"""Called when recording starts."""
pass
def on_recording_stop(self):
"""Called when recording stops."""
pass
def on_wakeword_detected(self):
"""Called when wake word is detected."""
pass
def on_wakeword_timeout(self):
"""Called when wake word times out."""
pass
def on_wakeword_detection_start(self):
"""Called when starting to listen for wake words."""
pass
def show_transcriptions(self, _):
"""Show recent transcriptions in an alert."""
if self.output_file and self.save_to_file and self.save_to_file.exists():
try:
with open(self.save_to_file, 'r', encoding='utf-8') as f:
lines = f.readlines()
recent = ''.join(lines[-10:]) # Last 10 lines
if recent.strip():
rumps.alert("Recent Transcriptions", recent)
self.is_paused = False
self.update_menu_states()
rumps.notification("STT Resumed", "Speech recognition resumed", f"Listening for '{self.wake_word}'")
def transcription_loop(self):
"""Main transcription loop running in background thread."""
while self.is_running and not self.stop_event.is_set():
try:
if not self.is_paused and self.recorder:
text = self.recorder.text()
if text and text.strip():
self.on_transcription_complete(text)
else:
rumps.alert("No Transcriptions", "No transcriptions found yet.")
time.sleep(0.1) # Small delay when paused
except Exception as e:
print(f"Transcription error: {e}")
time.sleep(1) # Longer delay on error
def on_realtime_transcription(self, text: str):
"""Handle real-time transcription updates."""
# Could update a notification or log
pass
def on_transcription_complete(self, text: str):
"""Handle completed transcriptions."""
if text.strip():
# Show notification with transcription
rumps.notification("Transcription", "Speech detected:", text[:100] + ("..." if len(text) > 100 else ""))
# Save to file if specified
if self.output_file:
timestamp = datetime.datetime.now().strftime("%H:%M:%S")
self.output_file.write(f"[{timestamp}] {text}\n")
self.output_file.flush()
def on_recording_start(self):
"""Called when recording starts."""
pass
def on_recording_stop(self):
"""Called when recording stops."""
pass
def on_wakeword_detected(self):
"""Called when wake word is detected."""
pass
def on_wakeword_timeout(self):
"""Called when wake word times out."""
pass
def on_wakeword_detection_start(self):
"""Called when starting to listen for wake words."""
pass
def show_transcriptions(self, _):
"""Show recent transcriptions in an alert."""
if self.output_file and self.save_to_file and self.save_to_file.exists():
try:
with open(self.save_to_file, 'r', encoding='utf-8') as f:
lines = f.readlines()
recent = ''.join(lines[-10:]) # Last 10 lines
if recent.strip():
rumps.alert("Recent Transcriptions", recent)
else:
rumps.alert("No Transcriptions", "No transcriptions found yet.")
except Exception as e:
rumps.alert("Error", f"Could not read transcriptions: {e}")
else:
rumps.alert("No File", "No output file configured. Use 'Save to File...' to set one.")
def select_output_file(self, _):
"""Select output file for saving transcriptions."""
try:
# Simple file selection using input dialog
response = rumps.Window(
message="Enter filename for transcriptions:",
title="Save Transcriptions",
default_text="stt_transcriptions.txt",
ok="Save",
cancel="Cancel"
).run()
if response.clicked and response.text:
self.save_to_file = Path.home() / "Documents" / response.text
rumps.notification("File Set", "Transcriptions will be saved to:", str(self.save_to_file))
except Exception as e:
rumps.alert("Error", f"Could not read transcriptions: {e}")
else:
rumps.alert("No File", "No output file configured. Use 'Save to File...' to set one.")
def select_output_file(self, _):
"""Select output file for saving transcriptions."""
try:
# Simple file selection using input dialog
response = rumps.Window(
message="Enter filename for transcriptions:",
title="Save Transcriptions",
default_text="stt_transcriptions.txt",
ok="Save",
cancel="Cancel"
).run()
if response.clicked and response.text:
self.save_to_file = Path.home() / "Documents" / response.text
rumps.notification("File Set", "Transcriptions will be saved to:", str(self.save_to_file))
except Exception as e:
rumps.alert("Error", f"Could not set output file: {e}")
rumps.alert("Error", f"Could not set output file: {e}")
@stt_app.command("listen")