def get_abspath(path):
"""Return the absolute path of <path>.
Because the working directory of MLHUB model is ~/.mlhub/<model>,
when user run 'ml score facematch <image-path>', the <image-path> may be a
path relative to the path where 'ml score facematch' is typed, to cope with
this scenario, mlhub provides mlhub.utils.get_cmd_cwd() to obtain this path.
"""
path = os.path.expanduser(path)
if not os.path.isabs(path):
path = os.path.join(get_cmd_cwd(), path)
return os.path.abspath(path)
headers = {'User-Agent': 'Mozilla/5.0'}
req = urllib.request.Request(url, headers=headers)
if urllib.request.urlopen(req).status == 200:
try:
analysis = client.analyze_image_by_domain(
domain, url, language)
except Exception as e:
catch_exception(e, url)
except urllib.error.URLError:
sys.exit("Error: The URL does not appear to exist. Please check.\n"
f"{url}")
else:
path = os.path.join(get_cmd_cwd(), url)
with open(path, 'rb') as fstream:
try:
analysis = client.analyze_image_by_domain_in_stream(
domain, fstream, language)
except Exception as e:
catch_exception(e, path)
for landmark in analysis.result["landmarks"]:
print('{},{}'.format(
round(landmark["confidence"], 2),
landmark["name"],
))
# Write results to stdout
# -----------------------------------------------------------------------
# Set up a speech configuration.
# -----------------------------------------------------------------------
speech_config = speechsdk.SpeechConfig(subscription=key, region=location)
if args.lang:
speech_config.speech_recognition_language = args.lang
# -----------------------------------------------------------------------
# Transcribe file or from microphone.
# -----------------------------------------------------------------------
if args.input:
path = os.path.join(get_cmd_cwd(), args.input)
if not os.path.exists(path):
sys.exit(f"azspeech transcribe: File not found: {path}")
try:
w = wave.open(path)
except Exception as e:
# print(e)
sys.exit(
f"azspeech transcribe: File does not seem to be wav audio: {path}")
# Create a callback to terminate the transcription once the full
# audio has been transcribed.
done = False
height = 50
if is_url(url):
try:
headers = {'User-Agent': 'Mozilla/5.0'}
req = urllib.request.Request(url, headers=headers)
if urllib.request.urlopen(req).status == 200:
try:
analysis = client.generate_thumbnail(width, height, url)
except Exception as e:
catch_exception(e, url)
sname = re.sub('\.(\w+)$', r'-thumbnail.\1',
os.path.basename(urlparse(url).path))
sname = os.path.join(get_cmd_cwd(), sname)
except urllib.error.URLError:
sys.exit("Error: The URL does not appear to exist. Please check."
f"{url}")
else:
path = os.path.join(get_cmd_cwd(), url)
with open(path, 'rb') as fstream:
try:
analysis = client.generate_thumbnail_in_stream(
width, height, fstream)
except Exception as e:
catch_exception(e, path)
sname = re.sub('\.(\w+)$', r'-thumbnail.\1', path)
def recognise(input_file, single_line, key):
# -----------------------------------------------------------------------
# Create verification voice profile
# -----------------------------------------------------------------------
create_headers = {
'Ocp-Apim-Subscription-Key': key,
'Content-Type': 'application/json'
}
endpoint = 'https://westus.api.cognitive.microsoft.com/'
path = 'speaker/verification/v2.0/text-dependent/profiles'
create_profile_url = endpoint + path
result = requests.post(create_profile_url,
data="{\"locale\":\"en-US\"}",
headers=create_headers)
# -----------------------------------------------------------------------
# Enroll the voice profile (3 samples)
# -----------------------------------------------------------------------
# If the location is not westus, profileId will not be found.
try:
profile_id = result.json()['profileId']
except:
if result.json()['error']['code'] == '401':
sys.exit(
"The Azure Speech key is not correct, please run ml configure azspeech to update your key. "
)
else:
error = result.json()['error']['message']
sys.exit(f"Error: {error}")
enroll_url = create_profile_url + "/" + profile_id + "/enrollments"
enroll_header = {
'Ocp-Apim-Subscription-Key': key,
'Content-Type': 'audio/wav; codecs=audio/pcm; samplerate=16000'
}
# Add three sample audios
for i in range(0, 3):
path = os.path.join(get_cmd_cwd(), input_file[i])
try:
w = wave.open(path, "rb")
except FileNotFoundError:
sys.exit(f"Error: wrong sample file location. \n{input_file[i]}")
# Convert audio file into binary format
binary_data = w.readframes(w.getnframes())
w.close()
result = requests.post(enroll_url,
data=binary_data,
headers=enroll_header)
# Catch the invalid audios
try:
if result.json()['error']['message']:
error = result.json()['error']['message']
print(f"The sample audio file {input_file[i]} error: {error}",
file=sys.stderr)
except:
pass
else:
sys.exit(1)
# -----------------------------------------------------------------------
# Verify the audio
# -----------------------------------------------------------------------
verify_url = create_profile_url + "/" + profile_id + "/verify"
verify_header = {
'Ocp-Apim-Subscription-Key': key,
'Content-Type': 'audio/wav; codecs=audio/pcm; samplerate=16000'
}
path = os.path.join(get_cmd_cwd(), input_file[3])
try:
w = wave.open(path, "rb")
except FileNotFoundError:
sys.exit(f"Error: wrong verification file location.\n{input_file[3]}")
# Convert audio file into binary format
binary_data = w.readframes(w.getnframes())
w.close()
result = requests.post(verify_url, data=binary_data, headers=verify_header)
try:
if not single_line:
print("Result: " + result.json()['recognitionResult'])
print("Score: " + str(result.json()['score']))
else:
print(result.json()['recognitionResult'] + ", " +
str(result.json()['score']))
except:
error = result.json()['error']['message']
sys.exit(f"Error: {error}")
# -----------------------------------------------------------------------
# Delete the voice profile
# -----------------------------------------------------------------------
delete_url = create_profile_url + "/" + profile_id
delete_header = {
'Ocp-Apim-Subscription-Key': key,
}
requests.delete(delete_url, headers=delete_header)
'Ocp-Apim-Subscription-Key': key,
'Ocp-Apim-Subscription-Region': location,
'Content-type': 'application/json',
}
endpoint = 'https://api.cognitive.microsofttranslator.com/'
path = '/translate?api-version=3.0'
translate_url = endpoint + path
# ----------------------------------------------------------------------
# Read the text to be translated.
# ----------------------------------------------------------------------
text = ""
if args.input:
text = open(os.path.join(get_cmd_cwd(), args.input), "r").read()
elif args.sentence:
text = " ".join(args.sentence)
# ----------------------------------------------------------------------
# Support function to translate the text.
# ----------------------------------------------------------------------
def translate(text, to):
json = [{'text': text}]
params = f'&to={to}'
result = requests.post(translate_url + params, headers=headers, json=json)
result = result.json()
sys.stdout.write(f"{result[0]['detectedLanguage']['language']}," +
url = os.path.abspath(os.path.expanduser(url))
if CMD_CWD != '':
os.chdir(oldwd)
if os.path.isdir(url):
for img in os.listdir(url):
img_file = os.path.join(url, '', img)
_colorize_one_img(img_file)
else:
_colorize_one_img(url)
# The working dir of the command which invokes this script.
CMD_CWD = get_cmd_cwd()
# Setup input path completion
readline.set_completer_delims('\t')
readline.parse_and_bind("tab: complete")
readline.set_completer(_tab_complete_path)
# Scoring
if len(args.path) == 0:
try:
url = input("Path or URL of images to colorize (Quit by Ctrl-d):\n(You could try images in '~/.mlhub/colorize/images/')\n> ")
except EOFError:
print()
help='path to an audio file to save. The file type should be wav')
args = option_parser.parse_args()
# ----------------------------------------------------------------------
# Request subscription key and location from user.
# ----------------------------------------------------------------------
key, location = get_private()
# ----------------------------------------------------------------------
# Read the text to be translated.
# ----------------------------------------------------------------------
text = ""
if args.input:
text = open(os.path.join(get_cmd_cwd(), args.input), "r").read()
elif args.sentence:
text = " ".join(args.sentence)
# Split the text into a list of sentences. Each sentence is sent off
# for synthesis. This avoids a very long text going off to the
# synthesizer (which seems to have a limit of some 640 characters) and
# is a natural break point anyhow.
text = " ".join(text.splitlines())
text = text.replace(". ", "\n")
text = text.splitlines()
# -----------------------------------------------------------------------
# Set up a speech synthesizer using the default speaker as audio output.
#
def synthesize_translations(to_language, single_line, result, output, key,
region):
if not single_line:
print(f'Recognized: "{result.text}"')
try:
translation = result.translations[to_language]
except Exception as e:
print(f"Error:{e}.", file=sys.stderr)
print(
"Error: wrong original or target language code. For original language code, please choose one from "
"Loale in Speech-to-text table"
"(https://docs.microsoft.com/en-us/azure/cognitive-services/speech-service/language-support#speech"
"-to-text).\n "
"For target language code, please choose one from Language Code in Text languages table"
"(https://docs.microsoft.com/en-us/azure/cognitive-services/speech-service/language-support#speech"
"-translation)",
file=sys.stderr)
sys.exit(1)
print(f'Translated into "{to_language}": {translation}')
else:
try:
translation = result.translations[to_language]
except Exception as e:
print(f"Error:{e}.", file=sys.stderr)
print(
"Error: wrong original or target language code. For original language code, please choose one from "
"Loale in Speech-to-text table"
"(https://docs.microsoft.com/en-us/azure/cognitive-services/speech-service/language-support#speech"
"-to-text).\n "
"For target language code, please choose one from Language Code in Text languages table"
"(https://docs.microsoft.com/en-us/azure/cognitive-services/speech-service/language-support#speech"
"-translation)",
file=sys.stderr)
sys.exit(1)
print(
f'Recognized: "{result.text}". Translated into "{to_language}": {translation}'
)
# ----------------------------------------------------------------------
# Get language code. "Original": is from Loale in Speech-to-text table.
# (https://docs.microsoft.com/en-us/azure/cognitive-services/speech-service/language-support#speech-to-text)
# "target": is from Language Code in Text languages table.
# (https://docs.microsoft.com/en-us/azure/cognitive-services/speech-service/language-support#speech-translation)
# Make sure the text language in Text Language table exists in Neural
# voices table.
# ----------------------------------------------------------------------
language_to_voice_map = {}
dataframe_speech = pandas.read_csv(os.path.join(os.getcwd(),
"data/text-to-speech.txt"),
delimiter="\t")
for index, row in dataframe_speech.iterrows():
if row[1] == "zh-HK":
language_code = "yue"
elif row[1] == "zh-CN":
language_code = "zh-Hant"
else:
language_code = row[1][0:2]
language_to_voice_map[language_code] = row[3]
speech_conf = speechsdk.SpeechConfig(subscription=key, region=region)
try:
speech_conf.speech_synthesis_voice_name = language_to_voice_map.get(
to_language)
except:
print(
f"Error: This target language ({to_language}) doesn't have speech.",
file=sys.stderr)
sys.exit(1)
if output:
file_location = os.path.join(get_cmd_cwd(), output)
audio_conf = speechsdk.audio.AudioOutputConfig(filename=file_location)
synthesizer = speechsdk.SpeechSynthesizer(speech_config=speech_conf,
audio_config=audio_conf)
else:
synthesizer = speechsdk.SpeechSynthesizer(speech_config=speech_conf)
synthesizer.speak_text_async(translation).get()