def pronunciation_assessment_from_microphone():
""""performs one-shot pronunciation assessment asynchronously with input from microphone."""
# Creates an instance of a speech config with specified subscription key and service region.
# Replace with your own subscription key and service region (e.g., "westus").
# Note: The pronunciation assessment feature is currently only available on westus, eastasia and centralindia regions.
# And this feature is currently only available on en-US language.
config = speechsdk.SpeechConfig(subscription=speech_key, region=service_region)
reference_text = ""
# create pronunciation assessment config, set grading system, granularity and if enable miscue based on your requirement.
pronunciation_config = speechsdk.PronunciationAssessmentConfig(reference_text=reference_text,
grading_system=speechsdk.PronunciationAssessmentGradingSystem.HundredMark,
granularity=speechsdk.PronunciationAssessmentGranularity.Phoneme,
enable_miscue=True)
recognizer = speechsdk.SpeechRecognizer(speech_config=config)
while True:
# Receives reference text from console input.
print('Enter reference text you want to assess, or enter empty text to exit.')
print('> ')
try:
reference_text = input()
except EOFError:
break
pronunciation_config.reference_text = reference_text
pronunciation_config.apply_to(recognizer)
# Starts recognizing.
print('Read out "{}" for pronunciation assessment ...'.format(reference_text))
# Note: Since recognize_once() returns only a single utterance, it is suitable only for single
# shot evaluation.
# For long-running multi-utterance pronunciation evaluation, use start_continuous_recognition() instead.
result = recognizer.recognize_once_async().get()
# Check the result
if result.reason == speechsdk.ResultReason.RecognizedSpeech:
print('Recognized: {}'.format(result.text))
print(' Pronunciation Assessment Result:')
pronunciation_result = speechsdk.PronunciationAssessmentResult(result)
print(' Accuracy score: {}, Pronunciation score: {}, Completeness score : {}, FluencyScore: {}'.format(
pronunciation_result.accuracy_score, pronunciation_result.pronunciation_score,
pronunciation_result.completeness_score, pronunciation_result.fluency_score
))
print(' Word-level details:')
for idx, word in enumerate(pronunciation_result.words):
print(' {}: word: {}, accuracy score: {}, error type: {};'.format(
idx + 1, word.word, word.accuracy_score, word.error_type
))
elif result.reason == speechsdk.ResultReason.NoMatch:
print("No speech could be recognized")
elif result.reason == speechsdk.ResultReason.Canceled:
cancellation_details = result.cancellation_details
print("Speech Recognition canceled: {}".format(cancellation_details.reason))
if cancellation_details.reason == speechsdk.CancellationReason.Error:
print("Error details: {}".format(cancellation_details.error_details))
def recognized(evt):
print('pronunciation assessment for: {}'.format(evt.result.text))
pronunciation_result = speechsdk.PronunciationAssessmentResult(
evt.result)
print(
' Accuracy score: {}, pronunciation score: {}, completeness score : {}, fluency score: {}'
.format(pronunciation_result.accuracy_score,
pronunciation_result.pronunciation_score,
pronunciation_result.completeness_score,
pronunciation_result.fluency_score))
nonlocal recognized_words, accuracy_scores, durations, valid_durations, start_offset, end_offset
recognized_words += pronunciation_result.words
accuracy_scores.append(pronunciation_result.accuracy_score)
json_result = evt.result.properties.get(
speechsdk.PropertyId.SpeechServiceResponse_JsonResult)
jo = json.loads(json_result)
nb = jo['NBest'][0]
durations.append(sum([int(w['Duration']) for w in nb['Words']]))
if start_offset is None:
start_offset = nb['Words'][0]['Offset']
end_offset = nb['Words'][-1]['Offset'] + nb['Words'][-1][
'Duration'] + 100000
for w, d in zip(pronunciation_result.words, nb['Words']):
if w.error_type == 'None':
valid_durations.append(d['Duration'] + 100000)
def recognized(evt):
print('pronunciation assessment for: {}'.format(evt.result.text))
pronunciation_result = speechsdk.PronunciationAssessmentResult(evt.result)
print(' Accuracy score: {}, Pronunciation score: {}, Completeness score : {}, FluencyScore: {}'.format(
pronunciation_result.accuracy_score, pronunciation_result.pronunciation_score,
pronunciation_result.completeness_score, pronunciation_result.fluency_score
))
nonlocal recognized_words
recognized_words += pronunciation_result.words
def recognized(evt):
print('pronunciation assessment for: {}'.format(evt.result.text))
pronunciation_result = speechsdk.PronunciationAssessmentResult(
evt.result)
print(
' Accuracy score: {}, pronunciation score: {}, completeness score : {}, fluency score: {}'
.format(pronunciation_result.accuracy_score,
pronunciation_result.pronunciation_score,
pronunciation_result.completeness_score,
pronunciation_result.fluency_score))
nonlocal recognized_words, accuracy_scores, fluency_scores, durations
recognized_words += pronunciation_result.words
accuracy_scores.append(pronunciation_result.accuracy_score)
fluency_scores.append(pronunciation_result.fluency_score)
json_result = evt.result.properties.get(
speechsdk.PropertyId.SpeechServiceResponse_JsonResult)
jo = json.loads(json_result)
nb = jo['NBest'][0]
durations.append(sum([int(w['Duration']) for w in nb['Words']]))
def pronunciation_assessment_from_microphone():
""""
Performs one-shot pronunciation assessment asynchronously with input from microphone.
See more information at https://aka.ms/csspeech/pa
"""
# Creates an instance of a speech config with specified subscription key and service region.
# Replace with your own subscription key and service region (e.g., "westus").
config = speechsdk.SpeechConfig(subscription=speech_key, region=service_region)
# The pronunciation assessment service has a longer default end silence timeout (5 seconds) than normal STT
# as the pronunciation assessment is widely used in education scenario where kids have longer break in reading.
# You can adjust the end silence timeout based on your real scenario.
config.set_property(speechsdk.PropertyId.SpeechServiceConnection_EndSilenceTimeoutMs, "3000")
reference_text = ""
# create pronunciation assessment config, set grading system, granularity and if enable miscue based on your requirement.
pronunciation_config = speechsdk.PronunciationAssessmentConfig(
reference_text=reference_text,
grading_system=speechsdk.PronunciationAssessmentGradingSystem.HundredMark,
granularity=speechsdk.PronunciationAssessmentGranularity.Phoneme,
enable_miscue=True)
# Creates a speech recognizer, also specify the speech language
recognizer = speechsdk.SpeechRecognizer(speech_config=config, language="en-US")
while True:
# Receives reference text from console input.
print('Enter reference text you want to assess, or enter empty text to exit.')
print('> ')
try:
reference_text = input()
except EOFError:
break
pronunciation_config.reference_text = reference_text
pronunciation_config.apply_to(recognizer)
# Starts recognizing.
print('Read out "{}" for pronunciation assessment ...'.format(reference_text))
# Note: Since recognize_once() returns only a single utterance, it is suitable only for single
# shot evaluation.
# For long-running multi-utterance pronunciation evaluation, use start_continuous_recognition() instead.
result = recognizer.recognize_once_async().get()
# Check the result
if result.reason == speechsdk.ResultReason.RecognizedSpeech:
print('Recognized: {}'.format(result.text))
print(' Pronunciation Assessment Result:')
pronunciation_result = speechsdk.PronunciationAssessmentResult(result)
print(' Accuracy score: {}, Pronunciation score: {}, Completeness score : {}, FluencyScore: {}'.format(
pronunciation_result.accuracy_score, pronunciation_result.pronunciation_score,
pronunciation_result.completeness_score, pronunciation_result.fluency_score
))
print(' Word-level details:')
for idx, word in enumerate(pronunciation_result.words):
print(' {}: word: {}, accuracy score: {}, error type: {};'.format(
idx + 1, word.word, word.accuracy_score, word.error_type
))
elif result.reason == speechsdk.ResultReason.NoMatch:
print("No speech could be recognized")
elif result.reason == speechsdk.ResultReason.Canceled:
cancellation_details = result.cancellation_details
print("Speech Recognition canceled: {}".format(cancellation_details.reason))
if cancellation_details.reason == speechsdk.CancellationReason.Error:
print("Error details: {}".format(cancellation_details.error_details))