def cli_cmd_load_directory(args):
# Set the logging level.
_set_logging_level(args)
# Initialise the specified engine. Return early if there was an error.
engine = _init_engine(args)
if engine is None:
return 1
# Connect the engine, load each command module and start the main
# recognition loop. The loop will normally exit on engine.disconnect()
# or a keyboard interrupt.
LOG.debug("Recognizing with engine '%s'", args.engine)
with engine.connection():
if args.recursive:
LOG.info("Loading command modules in sub-directories as "
"specified (recursive mode).")
directory = CommandModuleDirectory(args.module_dir,
recursive=args.recursive)
directory.load()
return_code = 0 if directory.loaded else 1
# Return early if --no-input was specified.
if args.no_input:
return return_code
_do_recognition(engine, args)
# Return the success of module loading.
return return_code
def main():
logging.basicConfig(level=logging.INFO)
try:
path = os.path.dirname(__file__)
except NameError:
# The "__file__" name is not always available, for example
# when this module is run from PythonWin. In this case we
# simply use the current working directory.
path = os.getcwd()
__file__ = os.path.join(path, "wsr_module_loader_plus.py")
# Set any configuration options here as keyword arguments.
engine = get_engine("sapi5inproc")
# Call connect() now that the engine configuration is set.
engine.connect()
# Register a recognition observer
observer = Observer()
observer.register()
load_sleep_wake_grammar(True)
directory = CommandModuleDirectory(path, excludes=[__file__])
directory.load()
# Start the engine's main recognition loop
try:
# Recognize from WSR in a loop.
print("Listening...")
engine.recognize_forever()
except KeyboardInterrupt:
pass
def main():
try:
path = os.path.dirname(__file__)
except NameError:
# The "__file__" name is not always available, for example
# when this module is run from PythonWin. In this case we
# simply use the current working directory.
path = os.getcwd()
__file__ = os.path.join(path, "dfly-loader-wsr.py")
# Initialize and connect the engine.
# Set any configuration options here as keyword arguments.
engine = get_engine("sapi5inproc")
engine.connect()
# Load grammars.
directory = CommandModuleDirectory(path, excludes=[__file__])
directory.load()
# Define recognition callback functions.
def on_begin():
print("Speech start detected.")
def on_recognition(words):
print("Recognized: %s" % " ".join(words))
def on_failure():
print("Sorry, what was that?")
# Recognize from WSR in a loop.
try:
engine.do_recognition(on_begin, on_recognition, on_failure)
except KeyboardInterrupt:
pass
def main():
try:
path = os.path.dirname(__file__)
except NameError:
# The "__file__" name is not always available, for example
# when this module is run from PythonWin. In this case we
# simply use the current working directory.
path = os.getcwd()
__file__ = os.path.join(path, "dfly-loader-natlink.py")
engine = get_engine("natlink")
engine.connect()
# Register a recognition observer
observer = Observer()
observer.register()
directory = CommandModuleDirectory(path, excludes=[__file__])
directory.load()
# Recognize from Dragon in a loop (opens a dialogue window).
engine.natlink.waitForSpeech()
# Unload all grammars from the engine so that Dragon doesn't keep
# recognizing them.
for grammar in engine.grammars:
grammar.unload()
# Disconnect after the dialogue is closed.
engine.disconnect()
def main():
logging.basicConfig(level=logging.INFO)
try:
path = os.path.dirname(__file__)
except NameError:
# The "__file__" name is not always available, for example
# when this module is run from PythonWin. In this case we
# simply use the current working directory.
path = os.getcwd()
__file__ = os.path.join(path, "dfly-loader-wsr.py")
engine = get_engine("sapi5inproc")
engine.connect()
# Register a recognition observer
observer = Observer()
observer.register()
directory = CommandModuleDirectory(path, excludes=[__file__])
directory.load()
engine.speak('beginning loop!')
while 1:
pythoncom.PumpWaitingMessages()
time.sleep(.1)
def main():
try:
path = os.path.dirname(__file__)
except NameError:
# The "__file__" name is not always available, for example
# when this module is run from PythonWin. In this case we
# simply use the current working directory.
path = os.getcwd()
__file__ = os.path.join(path, "sphinx_module_loader.py")
# Initialize the engine.
engine = get_engine("sphinx")
# Try to import the local engine configuration object first. If there isn't one,
# use the default engine configuration.
log = logging.getLogger("config")
try:
import config
engine.config = config
log.info("Using local engine configuration module 'config.py'")
except ImportError:
pass
except Exception as e:
# Log errors caught when setting the configuration.
log.exception("Failed to set config using 'config.py': %s" % e)
log.warning("Falling back to the default engine configuration")
# You can also set any configuration options here instead of using a
# config.py file. For example:
engine.config.START_ASLEEP = False
engine.config.DECODER_CONFIG.set_float("-vad_threshold", 3.1)
engine.config.LANGUAGE = LANGUAGE
# Call connect() now that the engine configuration is set.
engine.connect()
# Load grammars.
directory = CommandModuleDirectory(path, excludes=[__file__])
directory.load()
class Observer(RecognitionObserver):
def on_begin(self):
print("=> Speech start detected.")
def on_recognition(self, words):
message = u"Recognized: %s" % u" ".join(words)
print("=>", message)
def on_failure(self):
print("=> Sorry, what was that?")
observer = Observer()
observer.register()
# Start the engine's main recognition loop
try:
engine.recognise_forever()
except KeyboardInterrupt:
pass
def main():
try:
path = os.path.dirname(__file__)
except NameError:
# The "__file__" name is not always available, for example
# when this module is run from PythonWin. In this case we
# simply use the current working directory.
path = os.getcwd()
__file__ = os.path.join(path, "sphinx_module_loader.py")
# Initialize the engine.
engine = get_engine("sphinx")
# Try to import the local engine configuration object first. If there isn't one,
# use the default engine configuration.
log = logging.getLogger("config")
try:
import config
engine.config = config
log.info("Using local engine configuration module 'config.py'")
except ImportError:
pass
except Exception as e:
# Log errors caught when setting the configuration.
log.exception("Failed to set config using 'config.py': %s" % e)
log.warning("Falling back to the default engine configuration")
# You can also set any configuration options here instead of using a
# config.py file. For example:
# engine.config.START_ASLEEP = False
# Call connect() now that the engine configuration is set.
engine.connect()
# Load grammars.
directory = CommandModuleDirectory(path, excludes=[__file__])
directory.load()
# Define recognition callback functions.
def on_begin():
print("Speech start detected.")
def on_recognition(words):
message = u"Recognized: %s" % u" ".join(words)
# This only seems to be an issue with Python 2.7 on Windows.
if six.PY2:
encoding = sys.stdout.encoding or "ascii"
message = message.encode(encoding, errors='replace')
print(message)
def on_failure():
print("Sorry, what was that?")
# Start the engine's main recognition loop
try:
engine.do_recognition(on_begin, on_recognition, on_failure)
except KeyboardInterrupt:
pass
def main():
logging.basicConfig(level=logging.INFO)
try:
path = os.path.dirname(__file__)
except NameError:
# The "__file__" name is not always available, for example
# when this module is run from PythonWin. In this case we
# simply use the current working directory.
path = os.getcwd()
__file__ = os.path.join(path, "module_loader.py")
# Set any configuration options here as keyword arguments.
# See Kaldi engine documentation for all available options and more info.
engine = get_engine(
'kaldi',
model_dir='kaldi_model', # default model directory
# vad_aggressiveness=3, # default aggressiveness of VAD
# vad_padding_start_ms=150, # default ms of required silence before VAD
# vad_padding_end_ms=150, # default ms of required silence after VAD
# vad_complex_padding_end_ms=500, # default ms of required silence after VAD for complex utterances
# input_device_index=None, # set to an int to choose a non-default microphone
# lazy_compilation=True, # set to True to parallelize & speed up loading
# retain_dir=None, # set to a writable directory path to retain recognition metadata and/or audio data
# retain_audio=None, # set to True to retain speech data wave files in the retain_dir (if set)
)
# Call connect() now that the engine configuration is set.
engine.connect()
# Load grammars.
load_sleep_wake_grammar(True)
directory = CommandModuleDirectory(path, excludes=[__file__])
directory.load()
# Define recognition callback functions.
def on_begin():
print("Speech start detected.")
def on_recognition(words):
message = u"Recognized: %s" % u" ".join(words)
# This only seems to be an issue with Python 2.7 on Windows.
if six.PY2:
encoding = sys.stdout.encoding or "ascii"
message = message.encode(encoding, errors='replace')
print(message)
def on_failure():
print("Sorry, what was that?")
# Start the engine's main recognition loop
engine.prepare_for_recognition()
try:
print("Listening...")
engine.do_recognition(on_begin, on_recognition, on_failure)
except KeyboardInterrupt:
pass
def main():
logging.basicConfig(level=logging.INFO)
try:
path = os.path.dirname(__file__)
except NameError:
# The "__file__" name is not always available, for example
# when this module is run from PythonWin. In this case we
# simply use the current working directory.
path = os.getcwd()
__file__ = os.path.join(path, "dfly-loader-wsr.py")
engine = get_engine("sapi5inproc")
engine.connect()
# Register a recognition observer
observer = Observer()
observer.register()
directory = CommandModuleDirectory(path, excludes=[__file__])
directory.load()
WinEventProcType = ctypes.WINFUNCTYPE(
None, ctypes.wintypes.HANDLE, ctypes.wintypes.DWORD,
ctypes.wintypes.HWND, ctypes.wintypes.LONG, ctypes.wintypes.LONG,
ctypes.wintypes.DWORD, ctypes.wintypes.DWORD)
def callback(hWinEventHook, event, hwnd, idObject, idChild, dwEventThread,
dwmsEventTime):
window = Window.get_foreground()
if hwnd == window.handle:
for grammar in engine.grammars:
# Prevent 'notify_begin()' from being called.
if grammar.name == "_recobs_grammar":
continue
grammar.process_begin(window.executable, window.title,
window.handle)
def set_hook(win_event_proc, event_type):
return ctypes.windll.user32.SetWinEventHook(
event_type, event_type, 0, win_event_proc, 0, 0,
win32con.WINEVENT_OUTOFCONTEXT)
win_event_proc = WinEventProcType(callback)
ctypes.windll.user32.SetWinEventHook.restype = ctypes.wintypes.HANDLE
[
set_hook(win_event_proc, et) for et in {
win32con.EVENT_SYSTEM_FOREGROUND,
win32con.EVENT_OBJECT_NAMECHANGE,
}
]
engine.speak('beginning loop!')
while 1:
pythoncom.PumpWaitingMessages()
time.sleep(.1)
def main():
logging.basicConfig(level=logging.INFO)
try:
path = os.path.dirname(__file__)
except NameError:
# The "__file__" name is not always available, for example
# when this module is run from PythonWin. In this case we
# simply use the current working directory.
path = os.getcwd()
__file__ = os.path.join(path, "kaldi_module_loader_plus.py")
if 'devices' in sys.argv:
print('microphones:', get_engine('kaldi').print_mic_list())
exit(0)
# Set any configuration options here as keyword arguments.
if 'test' in sys.argv:
engine = get_engine('text')
else:
engine = get_engine(
"kaldi",
model_dir='kaldi_model',
# tmp_dir='kaldi_model_zamia.tmp', # default for temporary directory
# vad_aggressiveness=3, # default aggressiveness of VAD
# vad_padding_start_ms=300, # default ms of required silence before VAD
vad_padding_end_ms=500, # default ms of required silence after VAD
# vad_complex_padding_end_ms=500, # default ms of required silence after VAD for complex utterances
audio_input_device='Blue',
# auto_add_to_user_lexicon=True, # set to True to possibly use cloud for pronunciations
# lazy_compilation=True, # set to True to parallelize & speed up loading
# cloud_dictation=None, # set to 'gcloud' to use cloud dictation
)
# Call connect() now that the engine configuration is set.
engine.connect()
# Register a recognition observer
observer = Observer()
observer.register()
load_sleep_wake_grammar(True)
# load_noise_grammar()
directory = CommandModuleDirectory(path, excludes=[__file__])
directory.load()
# Start the engine's main recognition loop
if 'test' not in sys.argv:
engine.prepare_for_recognition()
try:
# Loop forever
print("Listening...")
engine.do_recognition()
except KeyboardInterrupt:
pass
def main(args):
logging.basicConfig(level=logging.INFO)
try:
path = os.path.dirname(__file__)
except NameError:
# The "__file__" name is not always available, for example
# when this module is run from PythonWin. In this case we
# simply use the current working directory.
path = os.getcwd()
__file__ = os.path.join(path, "kaldi_module_loader_plus.py")
os.environ.setdefault("GOOGLE_APPLICATION_CREDENTIALS", "gcloud.json")
# Set any configuration options here as keyword arguments.
engine = get_engine(
"kaldi",
model_dir='kaldi_model_zamia',
# tmp_dir='kaldi_tmp', # default for temporary directory
# vad_aggressiveness=3, # default aggressiveness of VAD
# vad_padding_ms=300,
#vad_padding_start_ms=300, # default ms of required silence before VAD
#vad_padding_end_ms=300, # default ms of required silence after VAD
#vad_complex_padding_end_ms=1000, # default ms of required silence after VAD for complex utterances
# input_device_index=None, # set to an int to choose a non-default microphone
auto_add_to_user_lexicon=
True, # set to True to possibly use cloud for pronunciations
cloud_dictation_lang="en-US", # set to 'gcloud' to use cloud dictation
)
if len(args) >= 1 and args[0] == "-l":
# Show the list of attached microphone devices.
# Note that this code should only be called after the engine has been initialized above with configuration options,
# otherwise if this line of code happens before the engine is created, get_engine will create the engine with default arguments.
get_engine("kaldi").print_mic_list()
# Call connect() now that the engine configuration is set.
engine.connect()
# Register a recognition observer
observer = Observer()
observer.register()
load_sleep_wake_grammar(True)
directory = CommandModuleDirectory(path, excludes=[__file__])
directory.load()
# Start the engine's main recognition loop
try:
# Loop forever
print("Listening...")
engine.do_recognition()
except KeyboardInterrupt:
pass
def _load_cmd_module_dirs(args):
# Load command modules from each specified directory. Errors during
# loading will be caught and logged.
recursive = args.recursive
return_code = 0
for d in args.module_dirs:
module_directory = CommandModuleDirectory(d, recursive=recursive)
module_directory.load()
if not module_directory.loaded:
return_code = 1
# Return the overall success of module loading.
return return_code
def main():
logging.basicConfig(level=logging.INFO)
try:
path = os.path.dirname(__file__)
except NameError:
# The "__file__" name is not always available, for example
# when this module is run from PythonWin. In this case we
# simply use the current working directory.
path = os.getcwd()
__file__ = os.path.join(path, "sphinx_module_loader.py")
engine = get_engine("sphinx")
assert isinstance(engine, SphinxEngine)
# Try to import the local engine configuration object first. If there isn't one,
# use the default engine configuration.
log = logging.getLogger("config")
try:
import config
engine.config = config
log.info("Using local engine configuration module 'config.py'")
except ImportError:
pass
except EngineError as e:
# Log EngineErrors caught when setting the configuration.
log.warning(e)
log.warning("Falling back to using the default engine configuration "
"instead of 'config.py'")
# Call connect() now that the engine configuration is set.
engine.connect()
# Register a recognition observer
observer = Observer(engine)
observer.register()
directory = CommandModuleDirectory(path, excludes=[__file__])
directory.load()
# TODO Have a simple GUI for pausing, resuming, cancelling and stopping recognition, etc
# TODO Change script to import all modules before loading the grammars into Pocket Sphinx
# Start the engine's main recognition loop
try:
engine.recognise_forever()
except KeyboardInterrupt:
pass
def main():
logging.basicConfig(level=logging.INFO)
try:
path = os.path.dirname(__file__)
except NameError:
# The "__file__" name is not always available, for example
# when this module is run from PythonWin. In this case we
# simply use the current working directory.
path = os.getcwd()
__file__ = os.path.join(path, "sphinx_module_loader.py")
engine = get_engine("sphinx")
# Try to import the local engine configuration object first. If there isn't one,
# use the default engine configuration.
log = logging.getLogger("config")
try:
import config
engine.config = config
log.info("Using local engine configuration module 'config.py'")
except ImportError:
pass
except Exception as e:
# Log errors caught when setting the configuration.
log.exception("Failed to set config using 'config.py': %s" % e)
log.warning("Falling back to the default engine configuration")
# You can also set any configuration options here instead of using a
# config.py file. For example:
# engine.config.START_ASLEEP = False
# Call connect() now that the engine configuration is set.
engine.connect()
# Register a recognition observer
observer = Observer()
observer.register()
directory = CommandModuleDirectory(path, excludes=[__file__])
directory.load()
# Start the engine's main recognition loop
try:
engine.recognise_forever()
except KeyboardInterrupt:
pass
def main(args):
logging.basicConfig(level=logging.INFO)
try:
path = os.path.dirname(__file__)
except NameError:
# The "__file__" name is not always available, for example
# when this module is run from PythonWin. In this case we
# simply use the current working directory.
path = os.getcwd()
__file__ = os.path.join(path, "kaldi_module_loader_plus.py")
# Set any configuration options here as keyword arguments.
engine = get_engine(
"kaldi",
# model_dir='kaldi_model',
# tmp_dir='kaldi_model.tmp', # default for temporary directory
# vad_aggressiveness=3, # default aggressiveness of VAD
# vad_padding_end_ms=300, # default ms of required silence surrounding VAD
# vad_complex_padding_end_ms=1200,
# input_device_index=None, # set to an int to choose a non-default microphone
# auto_add_to_user_lexicon=True, # set to True to possibly use cloud for pronunciations
# cloud_dictation=None, # set to 'gcloud' to use cloud dictation
)
# Call connect() now that the engine configuration is set.
engine.connect()
# Register a recognition observer
observer = Observer()
observer.register()
load_sleep_wake_grammar(True)
directory = CommandModuleDirectory(path, excludes=[__file__])
directory.load()
# Start the engine's main recognition loop
engine.mimic("start listening")
engine.prepare_for_recognition()
try:
# Loop forever
print("Listening...")
engine.do_recognition()
except KeyboardInterrupt:
pass
def main():
try:
path = os.path.dirname(__file__)
except NameError:
# The "__file__" name is not always available, for example
# when this module is run from PythonWin. In this case we
# simply use the current working directory.
path = os.getcwd()
__file__ = os.path.join(path, "sphinx_module_loader.py")
engine = get_engine("sphinx")
# Try to import the local engine configuration object first. If there isn't one,
# use the default engine configuration.
log = logging.getLogger("config")
try:
import config
engine.config = config
log.info("Using local engine configuration module 'config.py'")
except ImportError:
pass
except Exception as e:
# Log errors caught when setting the configuration.
log.exception("Failed to set config using 'config.py': %s" % e)
log.warning("Falling back to the default engine configuration")
# You can also set any configuration options here instead of using a
# config.py file. For example:
# engine.config.START_ASLEEP = False
# Call connect() now that the engine configuration is set.
engine.connect()
# Register a recognition observer
observer = Observer()
observer.register()
directory = CommandModuleDirectory(path, excludes=[__file__])
directory.load()
# Start the engine's main recognition loop
try:
engine.recognise_forever()
except KeyboardInterrupt:
pass
def main():
logging.basicConfig(level=logging.INFO)
try:
path = os.path.dirname(__file__)
except NameError:
# The "__file__" name is not always available, for example
# when this module is run from PythonWin. In this case we
# simply use the current working directory.
path = os.getcwd()
__file__ = os.path.join(path, "wsr_module_loader_plus.py")
# Initialize and connect the engine.
# Set any configuration options here as keyword arguments.
engine = get_engine("sapi5inproc")
engine.connect()
# Load grammars.
load_sleep_wake_grammar(True)
directory = CommandModuleDirectory(path, excludes=[__file__])
directory.load()
# Define recognition callback functions.
def on_begin():
print("Speech start detected.")
def on_recognition(words):
message = u"Recognized: %s" % u" ".join(words)
# This only seems to be an issue with Python 2.7 on Windows.
if six.PY2:
encoding = sys.stdout.encoding or "ascii"
message = message.encode(encoding, errors='replace')
print(message)
def on_failure():
print("Sorry, what was that?")
# Start the engine's main recognition loop
try:
print("Listening...")
engine.do_recognition(on_begin, on_recognition, on_failure)
except KeyboardInterrupt:
pass
def main():
try:
path = os.path.dirname(__file__)
except NameError:
# The "__file__" name is not always available, for example
# when this module is run from PythonWin. In this case we
# simply use the current working directory.
path = os.getcwd()
__file__ = os.path.join(path, "dfly-loader-natlink.py")
# Initialize and connect the engine.
engine = get_engine("natlink")
engine.connect()
# Load grammars.
directory = CommandModuleDirectory(path, excludes=[__file__])
directory.load()
# Define recognition callback functions.
def on_begin():
print("Speech start detected.")
def on_recognition(words):
message = u"Recognized: %s" % u" ".join(words)
# This only seems to be an issue with Python 2.7 on Windows.
if six.PY2:
encoding = sys.stdout.encoding or "ascii"
message = message.encode(encoding, errors='replace')
print(message)
def on_failure():
print("Sorry, what was that?")
# Recognize from Dragon in a loop (opens a dialogue window).
try:
engine.do_recognition(on_begin, on_recognition, on_failure)
except KeyboardInterrupt:
pass
# Disconnect after the dialogue is closed.
engine.disconnect()
def main():
try:
path = os.path.dirname(__file__)
except NameError:
# The "__file__" name is not always available, for example
# when this module is run from PythonWin. In this case we
# simply use the current working directory.
path = os.getcwd()
__file__ = os.path.join(path, "kaldi_module_loader_plus.py")
# Set any configuration options here as keyword arguments.
# See Kaldi engine documentation for all available options and more info.
engine = get_engine(
"kaldi",
model_dir=
"models/daanzu_20200328_1ep-mediumlm", # default model directory
vad_aggressiveness=0, # default aggressiveness of VAD
vad_padding_start_ms=10, # default ms of required silence before VAD
vad_padding_end_ms=10, # default ms of required silence after VAD
vad_complex_padding_end_ms=
10, # default ms of required silence after VAD for complex utterances
# input_device_index=None, # set to an int to choose a non-default microphone
lazy_compilation=True, # set to True to parallelize & speed up loading
# retain_dir=None, # set to a writable directory path to retain recognition metadata and/or audio data
# retain_audio=None, # set to True to retain speech data wave files in the retain_dir (if set)
)
ui = App(do_quit=engine.disconnect)
def notify_status(status: AppStatus):
if status == AppStatus.LOADING:
print("Loading...")
ui.set_status_line("Initializing...")
elif status == AppStatus.SLEEPING:
print("Sleeping...")
ui.set_status_line("Asleep...")
elif status == AppStatus.READY:
print("Awake...")
ui.set_status_line("Listening...")
else:
print(f"Unknown status! {status}")
notify_status(AppStatus.LOADING)
# Call connect() now that the engine configuration is set.
engine.connect()
# Load grammars.
load_sleep_wake_grammar(initial_awake=True, notify_status=notify_status)
load_ui_grammar(do_quit=lambda: engine.disconnect(),
do_restart=restart_process)
directory = CommandModuleDirectory(path, excludes=[__file__])
directory.load()
# Define recognition callback functions.
def on_begin():
ui.set_visual_context("last speech start", datetime.datetime.now())
last_utterances = []
def on_recognition(words):
s = " ".join(words)
if len(s):
ui.set_last_heard(f"Last heard: {s}")
last_utterances.append(s)
while len(last_utterances) > MAX_DISPLAYED_HISTORY:
del last_utterances[0]
ui.set_visual_context("History", "\n" + "\n".join(last_utterances))
print("Recognized: %s" % " ".join(words))
def on_failure():
ui.set_visual_context("last speech failure", datetime.datetime.now())
# Start the engine's main recognition loop
engine.prepare_for_recognition()
watchdog_observer = start_watchdog_observer(do_restart=restart_process)
try:
notify_status(AppStatus.READY)
engine.do_recognition(
begin_callback=on_begin,
recognition_callback=on_recognition,
failure_callback=on_failure,
end_callback=None,
post_recognition_callback=None,
)
except KeyboardInterrupt:
print(f"Received keyboard interrupt so quitting...")
# cleanup
if watchdog_observer:
watchdog_observer.stop()
watchdog_observer.join()