async def main() -> None:
global mic_stdin_running, mic_stdin_thread
# Parse command-line arguments
parser = argparse.ArgumentParser(description="Rhasspy")
parser.add_argument("--profile",
"-p",
required=True,
type=str,
help="Name of profile to use")
parser.add_argument(
"--system-profiles",
type=str,
help="Directory with base profile files (read only)",
default=os.path.join(os.getcwd(), "profiles"),
)
parser.add_argument(
"--user-profiles",
type=str,
help="Directory with user profile files (read/write)",
default=os.path.expanduser("~/.config/rhasspy/profiles"),
)
parser.add_argument(
"--set",
"-s",
nargs=2,
action="append",
help="Set a profile setting value",
default=[],
)
parser.add_argument("--debug",
action="store_true",
help="Print DEBUG log to console")
parser.add_argument(
"--no-check",
action="store_true",
help="Don't check profile for necessary files",
)
sub_parsers = parser.add_subparsers(dest="command")
sub_parsers.required = True
# info
info_parser = sub_parsers.add_parser("info", help="Profile information")
info_parser.add_argument("--defaults",
action="store_true",
help="Only print default settings")
sentences_parser = sub_parsers.add_parser(
"sentences", help="Print profile sentences.ini")
# validate
# validate_parser = sub_parsers.add_parser(
# "validate", help="Validate profile against schema"
# )
# wav2text
wav2text_parser = sub_parsers.add_parser(
"wav2text", help="WAV file to text transcription")
wav2text_parser.add_argument("wav_files",
nargs="*",
help="Paths to WAV files")
# text2intent
text2intent_parser = sub_parsers.add_parser("text2intent",
help="Text parsed to intent")
text2intent_parser.add_argument("sentences",
nargs="*",
help="Sentences to parse")
text2intent_parser.add_argument("--handle",
action="store_true",
help="Pass result to intent handler")
# wav2intent
wav2intent_parser = sub_parsers.add_parser(
"wav2intent", help="WAV file to parsed intent")
wav2intent_parser.add_argument("wav_files",
nargs="*",
help="Paths to WAV files")
wav2intent_parser.add_argument("--handle",
action="store_true",
help="Pass result to intent handler")
# train
train_parser = sub_parsers.add_parser("train", help="Re-train profile")
# record
# record_parser = sub_parsers.add_parser('record', help='Record test phrases for profile')
# record_parser.add_argument('--directory', help='Directory to write WAV files and intent JSON files')
# record-wake
# record_wake_parser = sub_parsers.add_parser('record-wake', help='Record wake word examples for profile')
# record_wake_parser.add_argument('--directory', help='Directory to write WAV files')
# record_wake_parser.add_argument('--negative', action='store_true', help='Record negative examples (not the wake word)')
# tune
# tune_parser = sub_parsers.add_parser('tune', help='Tune speech acoustic model for profile')
# tune_parser.add_argument('--directory', help='Directory with WAV files and intent JSON files')
# tune-wake
# tune_wake_parser = sub_parsers.add_parser('tune-wake', help='Tune wake acoustic model for profile')
# tune_wake_parser.add_argument('--directory', help='Directory with WAV files')
# test
# test_parser = sub_parsers.add_parser('test', help='Test speech/intent recognizers for profile')
# test_parser.add_argument('directory', help='Directory with WAV files and intent JSON files')
# test-wake
# test_wake_parser = sub_parsers.add_parser(
# "test-wake", help="Test wake word examples for profile"
# )
# test_wake_parser.add_argument("directory", help="Directory with WAV files")
# test_wake_parser.add_argument(
# "--threads", type=int, default=4, help="Number of threads to use"
# )
# test_wake_parser.add_argument(
# "--system", type=str, default=None, help="Override wake word system"
# )
# mic2wav
mic2wav_parser = sub_parsers.add_parser("mic2wav",
help="Voice command to WAV data")
mic2wav_parser.add_argument(
"--timeout",
type=float,
default=None,
help="Maximum number of seconds to record (default=profile)",
)
# mic2text
mic2text_parser = sub_parsers.add_parser(
"mic2text", help="Voice command to text transcription")
mic2text_parser.add_argument(
"--timeout",
type=float,
default=None,
help="Maximum number of seconds to record (default=profile)",
)
# mic2intent
mic2intent_parser = sub_parsers.add_parser(
"mic2intent", help="Voice command to parsed intent")
mic2intent_parser.add_argument("--stdin",
action="store_true",
help="Read audio data from stdin")
mic2intent_parser.add_argument("--handle",
action="store_true",
help="Pass result to intent handler")
mic2intent_parser.add_argument(
"--timeout",
type=float,
default=None,
help="Maximum number of seconds to record (default=profile)",
)
# word2phonemes
word2phonemes_parser = sub_parsers.add_parser(
"word2phonemes", help="Get pronunciation(s) for word(s)")
word2phonemes_parser.add_argument("words",
nargs="*",
help="Word(s) to pronounce")
word2phonemes_parser.add_argument("-n",
type=int,
default=1,
help="Maximum number of pronunciations")
# word2wav
word2wav_parser = sub_parsers.add_parser("word2wav", help="Pronounce word")
word2wav_parser.add_argument("word", help="Word to pronounce")
# wav2mqtt
wav2mqtt_parser = sub_parsers.add_parser("wav2mqtt",
help="Push WAV file(s) to MQTT")
wav2mqtt_parser.add_argument("wav_files",
nargs="*",
help="Paths to WAV files")
wav2mqtt_parser.add_argument(
"--frames",
type=int,
default=480,
help="WAV frames per MQTT message (default=0 for all)",
)
wav2mqtt_parser.add_argument("--site-id",
type=str,
default="default",
help="Hermes siteId (default=default)")
wav2mqtt_parser.add_argument(
"--silence-before",
type=float,
default=0,
help="Seconds of silence to add before each WAV",
)
wav2mqtt_parser.add_argument(
"--silence-after",
type=float,
default=0,
help="Seconds of silence to add after each WAV",
)
wav2mqtt_parser.add_argument(
"--pause",
type=float,
default=0.01,
help="Seconds to wait before sending next chunk (default=0.01)",
)
# text2wav
text2wav_parser = sub_parsers.add_parser(
"text2wav", help="Output WAV file using text to speech system")
text2wav_parser.add_argument("sentence", help="Sentence to speak")
# text2speech
text2speech_parser = sub_parsers.add_parser(
"text2speech", help="Speak sentences using text to speech system")
text2speech_parser.add_argument("sentences",
nargs="*",
help="Sentences to speak")
# sleep
sleep_parser = sub_parsers.add_parser("sleep", help="Wait for wake word")
# download
download_parser = sub_parsers.add_parser("download",
help="Download profile files")
download_parser.add_argument(
"--delete",
action="store_true",
help="Clear download cache before downloading")
# check
check_parser = sub_parsers.add_parser(
"check", help="Check downloaded profile files")
# -------------------------------------------------------------------------
args = parser.parse_args()
if args.debug:
logging.root.setLevel(logging.DEBUG)
profiles_dirs = [args.system_profiles, args.user_profiles]
logger.debug(profiles_dirs)
default_settings = Profile.load_defaults(args.system_profiles)
# Create rhasspy core
from rhasspy.core import RhasspyCore
core = RhasspyCore(args.profile, args.system_profiles, args.user_profiles)
# Add profile settings from the command line
extra_settings = {}
for key, value in args.set:
try:
value = json.loads(value)
except:
pass
logger.debug("Profile: {0}={1}".format(key, value))
extra_settings[key] = value
core.profile.set(key, value)
# Handle command
if args.command == "info":
if args.defaults:
# Print default settings
json.dump(core.defaults, sys.stdout, indent=4)
else:
# Print profile settings
json.dump(core.profile.json, sys.stdout, indent=4)
# elif args.command == "validate":
# from cerberus import Validator
# schema_path = os.path.join(os.path.dirname(__file__), "profile_schema.json")
# with open(schema_path, "r") as schema_file:
# v = Validator(json.load(schema_file))
# if v.validate(core.profile.json):
# print("VALID")
# else:
# print("INVALID")
# for err in v._errors:
# print(err)
elif args.command == "sentences":
sentences_path = core.profile.read_path(
core.profile.get("speech_to_text.sentences_ini", "sentences.ini"))
with open(sentences_path, "r") as sentences_file:
sys.stdout.write(sentences_file.read())
else:
# Patch profile
profile = core.profile
profile.set("rhasspy.listen_on_start", False)
profile.set("rhasspy.preload_profile", False)
if args.command == "wav2mqtt":
profile.set("mqtt.enabled", True)
elif args.command in ["mic2intent"] and args.stdin:
profile.set("microphone.system", "stdin")
profile.set("microphone.stdin.auto_start", False)
mic_stdin_running = True
elif args.command == "text2wav":
profile.set("sounds.system", "dummy")
# Set environment variables
os.environ["RHASSPY_BASE_DIR"] = os.getcwd()
os.environ["RHASSPY_PROFILE"] = core.profile.name
os.environ["RHASSPY_PROFILE_DIR"] = core.profile.write_dir()
# Execute command
command_funcs = {
"wav2text": wav2text,
"text2intent": text2intent,
"wav2intent": wav2intent,
"train": train_profile,
# 'record': record,
# 'record-wake': record_wake,
# 'tune': tune,
# 'tune-wake': tune_wake,
# 'test': test,
# "test-wake": test_wake,
"mic2text": mic2text,
"mic2intent": mic2intent,
"mic2wav": mic2wav,
"word2phonemes": word2phonemes,
"word2wav": word2wav,
"wav2mqtt": wav2mqtt,
"text2wav": text2wav,
"text2speech": text2speech,
"sleep": sleep,
"download": download,
"check": check,
}
if not args.command in ["test-wake"]:
# Automatically start core
await core.start()
if not args.no_check and (args.command not in ["check", "download"]):
# Verify that profile has necessary files
missing_files = core.check_profile()
if len(missing_files) > 0:
logger.fatal(
f"Missing required files for {profile.name}: {missing_files.keys()}. Please run download command and try again."
)
sys.exit(1)
if mic_stdin_running:
logger.debug("Reading audio data from stdin")
mic_stdin_thread = threading.Thread(target=read_audio_stdin,
args=(core, ),
daemon=True)
mic_stdin_thread.start()
# Run command
try:
await command_funcs[args.command](core, profile, args)
if mic_stdin_thread is not None:
mic_stdin_running = False
mic_stdin_thread.join()
finally:
await core.shutdown()
async def check(core: RhasspyCore, profile: Profile, args: Any) -> None:
"""Verify that profile files are downloaded"""
missing_files = core.check_profile()
json.dump(missing_files, sys.stdout, indent=4)
async def check(core: RhasspyCore, profile: Profile, args: Any) -> None:
missing_files = core.check_profile()
json.dump(missing_files, sys.stdout, indent=4)
async def main() -> None:
"""Main method"""
global mic_stdin_running, mic_stdin_thread
# Parse command-line arguments
parser = argparse.ArgumentParser(description="Rhasspy")
parser.add_argument("--profile",
"-p",
required=True,
type=str,
help="Name of profile to use")
parser.add_argument(
"--system-profiles",
type=str,
help="Directory with base profile files (read only)",
default=os.path.join(os.getcwd(), "profiles"),
)
parser.add_argument(
"--user-profiles",
type=str,
help="Directory with user profile files (read/write)",
default=os.path.expanduser("~/.config/rhasspy/profiles"),
)
parser.add_argument(
"--set",
"-s",
nargs=2,
action="append",
help="Set a profile setting value",
default=[],
)
parser.add_argument("--debug",
action="store_true",
help="Print DEBUG log to console")
parser.add_argument(
"--no-check",
action="store_true",
help="Don't check profile for necessary files",
)
sub_parsers = parser.add_subparsers(dest="command")
sub_parsers.required = True
# info
info_parser = sub_parsers.add_parser("info", help="Profile information")
info_parser.add_argument("--defaults",
action="store_true",
help="Only print default settings")
# wav2text
wav2text_parser = sub_parsers.add_parser(
"wav2text", help="WAV file to text transcription")
wav2text_parser.add_argument("wav_files",
nargs="*",
help="Paths to WAV files")
# text2intent
text2intent_parser = sub_parsers.add_parser("text2intent",
help="Text parsed to intent")
text2intent_parser.add_argument("sentences",
nargs="*",
help="Sentences to parse")
text2intent_parser.add_argument("--handle",
action="store_true",
help="Pass result to intent handler")
# wav2intent
wav2intent_parser = sub_parsers.add_parser(
"wav2intent", help="WAV file to parsed intent")
wav2intent_parser.add_argument("wav_files",
nargs="*",
help="Paths to WAV files")
wav2intent_parser.add_argument("--handle",
action="store_true",
help="Pass result to intent handler")
# train
train_parser = sub_parsers.add_parser("train", help="Re-train profile")
train_parser.add_argument("--no-cache",
action="store_true",
help="Clear training cache")
# mic2wav
mic2wav_parser = sub_parsers.add_parser("mic2wav",
help="Voice command to WAV data")
mic2wav_parser.add_argument(
"--timeout",
type=float,
default=None,
help="Maximum number of seconds to record (default=profile)",
)
# mic2text
mic2text_parser = sub_parsers.add_parser(
"mic2text", help="Voice command to text transcription")
mic2text_parser.add_argument(
"--timeout",
type=float,
default=None,
help="Maximum number of seconds to record (default=profile)",
)
# mic2intent
mic2intent_parser = sub_parsers.add_parser(
"mic2intent", help="Voice command to parsed intent")
mic2intent_parser.add_argument("--stdin",
action="store_true",
help="Read audio data from stdin")
mic2intent_parser.add_argument("--handle",
action="store_true",
help="Pass result to intent handler")
mic2intent_parser.add_argument(
"--timeout",
type=float,
default=None,
help="Maximum number of seconds to record (default=profile)",
)
# word2phonemes
word2phonemes_parser = sub_parsers.add_parser(
"word2phonemes", help="Get pronunciation(s) for word(s)")
word2phonemes_parser.add_argument("words",
nargs="*",
help="Word(s) to pronounce")
word2phonemes_parser.add_argument("-n",
type=int,
default=1,
help="Maximum number of pronunciations")
# word2wav
word2wav_parser = sub_parsers.add_parser("word2wav", help="Pronounce word")
word2wav_parser.add_argument("word", help="Word to pronounce")
# wav2mqtt
wav2mqtt_parser = sub_parsers.add_parser("wav2mqtt",
help="Push WAV file(s) to MQTT")
wav2mqtt_parser.add_argument("wav_files",
nargs="*",
help="Paths to WAV files")
wav2mqtt_parser.add_argument(
"--frames",
type=int,
default=480,
help="WAV frames per MQTT message (default=0 for all)",
)
wav2mqtt_parser.add_argument("--site-id",
type=str,
default="default",
help="Hermes siteId (default=default)")
wav2mqtt_parser.add_argument(
"--silence-before",
type=float,
default=0,
help="Seconds of silence to add before each WAV",
)
wav2mqtt_parser.add_argument(
"--silence-after",
type=float,
default=0,
help="Seconds of silence to add after each WAV",
)
wav2mqtt_parser.add_argument(
"--pause",
type=float,
default=0.01,
help="Seconds to wait before sending next chunk (default=0.01)",
)
# text2wav
text2wav_parser = sub_parsers.add_parser(
"text2wav", help="Output WAV file using text to speech system")
text2wav_parser.add_argument("sentence", help="Sentence to speak")
# text2speech
text2speech_parser = sub_parsers.add_parser(
"text2speech", help="Speak sentences using text to speech system")
text2speech_parser.add_argument("sentences",
nargs="*",
help="Sentences to speak")
# sleep
sub_parsers.add_parser("sleep", help="Wait for wake word")
# download
download_parser = sub_parsers.add_parser("download",
help="Download profile files")
download_parser.add_argument(
"--delete",
action="store_true",
help="Clear download cache before downloading")
# check
sub_parsers.add_parser("check", help="Check downloaded profile files")
# -------------------------------------------------------------------------
args = parser.parse_args()
if args.debug:
logging.basicConfig(level=logging.DEBUG)
else:
logging.basicConfig(level=logging.INFO)
profiles_dirs = [args.system_profiles, args.user_profiles]
logger.debug(profiles_dirs)
# Create rhasspy core
core = RhasspyCore(args.profile, args.system_profiles, args.user_profiles)
# Add profile settings from the command line
extra_settings = {}
for key, value in args.set:
try:
value = json.loads(value)
except Exception:
pass
logger.debug("Profile: %s=%s", key, value)
extra_settings[key] = value
core.profile.set(key, value)
# Handle command
if args.command == "info":
if args.defaults:
# Print default settings
json.dump(core.defaults, sys.stdout, indent=4)
else:
# Print profile settings
json.dump(core.profile.json, sys.stdout, indent=4)
elif args.command == "sentences":
sentences_path = core.profile.read_path(
core.profile.get("speech_to_text.sentences_ini", "sentences.ini"))
with open(sentences_path, "r") as sentences_file:
sys.stdout.write(sentences_file.read())
else:
# Patch profile
profile = core.profile
profile.set("rhasspy.listen_on_start", False)
profile.set("rhasspy.preload_profile", False)
if args.command == "wav2mqtt":
profile.set("mqtt.enabled", True)
elif args.command in ["mic2intent"] and args.stdin:
profile.set("microphone.system", "stdin")
profile.set("microphone.stdin.auto_start", False)
mic_stdin_running = True
elif args.command == "text2wav":
profile.set("sounds.system", "dummy")
# Set environment variables
os.environ["RHASSPY_BASE_DIR"] = os.getcwd()
os.environ["RHASSPY_PROFILE"] = core.profile.name
os.environ["RHASSPY_PROFILE_DIR"] = core.profile.write_dir()
# Execute command
command_funcs = {
"wav2text": wav2text,
"text2intent": text2intent,
"wav2intent": wav2intent,
"train": train_profile,
"mic2text": mic2text,
"mic2intent": mic2intent,
"mic2wav": mic2wav,
"word2phonemes": word2phonemes,
"word2wav": word2wav,
"wav2mqtt": wav2mqtt,
"text2wav": text2wav,
"text2speech": text2speech,
"sleep": sleep,
"download": download,
"check": check,
}
# Automatically start core
await core.start()
if not args.no_check and (args.command not in ["check", "download"]):
# Verify that profile has necessary files
missing_files = core.check_profile()
if missing_files:
logger.fatal(
"Missing required files for %s: %s. Please run download command and try again.",
profile.name,
list(missing_files),
)
sys.exit(1)
if mic_stdin_running:
logger.debug("Reading audio data from stdin")
mic_stdin_thread = threading.Thread(target=read_audio_stdin,
args=(core, ),
daemon=True)
mic_stdin_thread.start()
# Run command
try:
await command_funcs[args.command](core, profile, args)
if mic_stdin_thread is not None:
mic_stdin_running = False
mic_stdin_thread.join()
finally:
await core.shutdown()
