def eval_error_textgrid_test(): audio_name = '05_Semahat_Ozdenses_-_Bir_Ihtimal_Daha_Var_0_zemin_from_69_5205_to_84_2' annotation_url = os.path.join(PATH_TEST_DATASET, audio_name + ANNOTATION_EXT) start_index = 0 end_index = -1 detected_token_list = [[0.61, 0.94, u'Bir'], [1.02, 3.41, u'ihtimal'], [3.42, 4.11, u'daha'], [4.12, 5.4, u'var'], [8.03, 8.42, u'o'], [8.46, 8.83, u'da'], [8.86, 10.65, u'\xf6lmek'], [10.66, 11.04, u'mi'], [11.05, 14.39, u'dersin']] annotation_token_list, detected_token_list, dummy, dummy = \ strip_non_lyrics_tokens(annotation_url, detected_token_list, tierAliases.phrases, start_index, end_index) alignment_errors = _eval_alignment_error(annotation_token_list, detected_token_list, tierAliases.phrases) mean, std_dev, median = getMeanAndStDevError(alignment_errors) print("mean : ", mean, "st dev: ", std_dev)
def eval_error_textgrid_test(): audio_name = '05_Semahat_Ozdenses_-_Bir_Ihtimal_Daha_Var_0_zemin_from_69_5205_to_84_2' annotation_url = os.path.join(PATH_TEST_DATASET, audio_name + ANNOTATION_EXT) start_index = 0 end_index = -1 detected_token_list = [[0.61, 0.94, u'Bir'], [1.02, 3.41, u'ihtimal'], [3.42, 4.11, u'daha'], [4.12, 5.4, u'var'], [8.03, 8.42, u'o'], [8.46, 8.83, u'da'], [8.86, 10.65, u'\xf6lmek'], [10.66, 11.04, u'mi'], [11.05, 14.39, u'dersin']] annotation_token_list, detected_token_list, dummy, dummy = \ strip_non_lyrics_tokens(annotation_url, detected_token_list, tierAliases.phrases, start_index, end_index) alignment_errors = _eval_alignment_error(annotation_token_list, detected_token_list, tierAliases.phrases) mean, std_dev, median = getMeanAndStDevError(alignment_errors) print("mean : ", mean, "st dev: ", std_dev)
def eval_percentage_correct_textgrid_test(): audio_name = '05_Semahat_Ozdenses_-_Bir_Ihtimal_Daha_Var_0_zemin_from_69_5205_to_84_2' annotation_url = os.path.join(PATH_TEST_DATASET, audio_name + ANNOTATION_EXT) detected_token_list = [[0.61, 0.94, u'Bir'], [1.02, 3.41, u'ihtimal'], [3.42, 4.11, u'daha'], [4.12, 5.4, u'var'], [8.03, 8.42, u'o'], [8.46, 8.83, u'da'], [8.86, 10.65, u'\xf6lmek'], [10.66, 11.04, u'mi'], [11.05, 14.39, u'dersin']] start_index = 0 end_index = -1 annotation_token_list, detected_token_list, final_ts_anno, initial_time_offset = \ strip_non_lyrics_tokens(annotation_url, detected_token_list, tierAliases.phrases, start_index, end_index) duration_correct, total_length = _eval_percentage_correct( annotation_token_list, detected_token_list, final_ts_anno, initial_time_offset) print(duration_correct / total_length)
def eval_percentage_correct_textgrid_test(): audio_name = '05_Semahat_Ozdenses_-_Bir_Ihtimal_Daha_Var_0_zemin_from_69_5205_to_84_2' annotation_url = os.path.join(PATH_TEST_DATASET, audio_name + ANNOTATION_EXT) detected_token_list = [[0.61, 0.94, u'Bir'], [1.02, 3.41, u'ihtimal'], [3.42, 4.11, u'daha'], [4.12, 5.4, u'var'], [8.03, 8.42, u'o'], [8.46, 8.83, u'da'], [8.86, 10.65, u'\xf6lmek'], [10.66, 11.04, u'mi'], [11.05, 14.39, u'dersin']] start_index = 0 end_index = -1 annotation_token_list, detected_token_list, final_ts_anno, initial_time_offset = \ strip_non_lyrics_tokens(annotation_url, detected_token_list, tierAliases.phrases, start_index, end_index) duration_correct, total_length = _eval_percentage_correct(annotation_token_list, detected_token_list, final_ts_anno, initial_time_offset) print(duration_correct / total_length)
def evalPercentageCorrect(annotationURI, outputHTKPhoneAlignedURI, whichTier, startIdx, endIdx ): ''' Wrapper around _evalPercentageCorrect() for txt file outputHTKPhoneAlignedURI Parameters -------------- outputHTKPhoneAlignedURI: detected timestamps in htk's mlf format other parameters same as in _evalPercentageCorrect ''' detectedTokenList = loadDetectedTokenListFromMlf( outputHTKPhoneAlignedURI, whichTier ) annotationTokenList, detectedTokenList, finalTsAnno, initialTimeOffset = \ strip_non_lyrics_tokens(annotationURI, detectedTokenList, whichTier, startIdx, endIdx) durationCorrect, totalDuration = _eval_percentage_correct(annotationTokenList, detectedTokenList, finalTsAnno, initialTimeOffset) return durationCorrect, totalDuration