Spaces:
Running
Running
alessandro trinca tornidor
feat: port whisper and faster-whisper support from https://github.com/Thiagohgl/ai-pronunciation-trainer
85b7206
import os | |
import platform | |
import unittest | |
import numpy as np | |
## permit to import from parent directory also in | |
import sys | |
from pathlib import Path | |
parent = Path(__file__).parent.parent | |
sys.path.append(str(parent)) | |
import WordMatching | |
from constants import app_logger | |
from tests import set_seed | |
from tests import constants_wordmatching as const | |
class TestWordMatching(unittest.TestCase): | |
def setUp(self): | |
if platform.system() == "Windows" or platform.system() == "Win32": | |
os.environ["PYTHONUTF8"] = "1" | |
os.environ["IS_TESTING"] = "TRUE" | |
def tearDown(self): | |
if platform.system() == "Windows" or platform.system() == "Win32" and "PYTHONUTF8" in os.environ: | |
del os.environ["PYTHONUTF8"] | |
del os.environ["IS_TESTING"] | |
def test_get_word_distance_matrix(self): | |
words_estimated = ["hello", "world"] | |
words_real = ["hello", "word"] | |
expected_matrix = np.array([[0., 5.], [4., 1.], [5., 4.]]) | |
result_matrix = WordMatching.get_word_distance_matrix(words_estimated, words_real) | |
np.testing.assert_array_equal(result_matrix, expected_matrix) | |
def test_get_best_path_from_distance_matrix(self): | |
for word_distance_matrix, expected_result_indices in const.get_best_path_from_distance_matrix_constants: | |
set_seed() | |
result_indices = WordMatching.get_best_path_from_distance_matrix(word_distance_matrix) | |
np.testing.assert_array_equal(result_indices, expected_result_indices) | |
def test_get_best_path_from_distance_matrix_with_inf_values(self): | |
set_seed() | |
try: | |
word_distance_matrix = np.array([[np.inf, 1, 2]]) | |
result_indices = WordMatching.get_best_path_from_distance_matrix(word_distance_matrix) | |
app_logger.info(f"result_indices0: {result_indices}, {result_indices.shape} .") | |
self.assertIsInstance(result_indices, np.ndarray) | |
self.assertEqual(result_indices.shape, (3,)) | |
self.assertGreater(result_indices[0], 0) | |
self.assertGreater(result_indices[1], 0) | |
self.assertEqual(result_indices[2], 0) | |
word_distance_matrix = np.array([[-1, np.inf, 3]]) | |
result_indices = WordMatching.get_best_path_from_distance_matrix(word_distance_matrix) | |
app_logger.info(f"result_indices1: {result_indices}, {result_indices.shape} .") | |
self.assertLess(result_indices[0], 0) | |
self.assertGreater(result_indices[1], 0) | |
self.assertEqual(result_indices[2], 0) | |
word_distance_matrix = np.array([[2, -1, np.inf]]) | |
result_indices = WordMatching.get_best_path_from_distance_matrix(word_distance_matrix) | |
app_logger.info(f"result_indices2: {result_indices}, {result_indices.shape} .") | |
self.assertGreater(result_indices[0], 0) | |
self.assertGreater(result_indices[1], -1) | |
self.assertEqual(result_indices[2], 0) | |
word_distance_matrix = np.array([[np.inf, 1, 2], [1, np.inf, 3], [2, 3, np.inf], [-1, -np.inf, 1]]) | |
result_indices = WordMatching.get_best_path_from_distance_matrix(word_distance_matrix) | |
app_logger.info(f"result_indices3: {result_indices}, {result_indices.shape} .") | |
self.assertGreater(result_indices[0], 0) | |
self.assertGreater(result_indices[1], 0) | |
self.assertEqual(result_indices[2], 0) | |
except AssertionError as ae: | |
app_logger.error("ae:") | |
app_logger.error(ae) | |
raise ae | |
def test_getWhichLettersWereTranscribedCorrectly(self): | |
real_word = "hello" | |
transcribed_word = [x for x in "hxllo"] | |
expected_result = [1, 0, 1, 1, 1] | |
result = WordMatching.getWhichLettersWereTranscribedCorrectly(real_word, transcribed_word) | |
self.assertEqual(result, expected_result) | |
def test_get_best_mapped_words_false(self): | |
words_estimated = ["hello", "world"] | |
words_real = ["hello", "word"] | |
expected_words = ["hello", "world"] | |
expected_indices = [0, 1] | |
result_words, result_indices = WordMatching.get_best_mapped_words(words_estimated, words_real, use_dtw=False) | |
self.assertEqual(result_words, expected_words) | |
self.assertEqual(result_indices, expected_indices) | |
expected_mapped_letters = ['e', 's', 's', 'e', 'n', '-'] | |
expected_mapped_words_indices = [np.int64(0), np.int64(1), np.int64(2), np.int64(3), np.int64(4), -1] | |
output_mapped_letters, output_mapped_words_indices = WordMatching.get_best_mapped_words("essen", "essen?", use_dtw=False) | |
assert output_mapped_letters == expected_mapped_letters | |
assert output_mapped_words_indices == expected_mapped_words_indices | |
def test_get_word_distance_matrix_with_empty_lists(self): | |
words_estimated = [] | |
words_real = [] | |
expected_matrix = np.arange(0).reshape((1, 0)) | |
result_matrix = WordMatching.get_word_distance_matrix(words_estimated, words_real) | |
np.testing.assert_array_equal(result_matrix, expected_matrix) | |
def test_get_word_distance_matrix_with_different_lengths(self): | |
words_estimated = ["hello"] | |
words_real = ["hello", "world"] | |
expected_matrix = np.array([[0., 4.], [5., 5.]]) | |
result_matrix = WordMatching.get_word_distance_matrix(words_estimated, words_real) | |
np.testing.assert_array_equal(result_matrix, expected_matrix) | |
def test_get_best_path_from_distance_matrix_with_empty_matrix_indexerror(self): | |
word_distance_matrix = np.array([]) | |
with self.assertRaises(IndexError): | |
try: | |
WordMatching.get_best_path_from_distance_matrix(word_distance_matrix) | |
except IndexError as e: | |
msg = "tuple index out of range" | |
assert msg in str(e) | |
raise e | |
def test_getWhichLettersWereTranscribedCorrectly_with_empty_strings(self): | |
real_word = "" | |
transcribed_word = [""] | |
expected_result = [] | |
result = WordMatching.getWhichLettersWereTranscribedCorrectly(real_word, transcribed_word) | |
self.assertEqual(result, expected_result) | |
def test_getWhichLettersWereTranscribedCorrectly_with_different_lengths(self): | |
real_word = "hello" | |
transcribed_word = [x for x in "hello oo"] | |
expected_result = [1, 1, 1, 1, 1] | |
result = WordMatching.getWhichLettersWereTranscribedCorrectly(real_word, transcribed_word) | |
self.assertEqual(result, expected_result) | |
def test_getWhichLettersWereTranscribedCorrectly_wrong_number_elements_mapped_letters(self): | |
word_real = "ich" | |
mapped_letters=['i', 'c', 'h', "z"] | |
is_letter_correct1 = WordMatching.getWhichLettersWereTranscribedCorrectly(word_real, mapped_letters) # , mapped_letters_indices) | |
self.assertEqual(is_letter_correct1, [1, 1, 1]) | |
def test_getWhichLettersWereTranscribedCorrectly_wrong_number_elements_mapped_letters(self): | |
word_real = "ichh" | |
mapped_letters=['i', 'c', 'h'] | |
with self.assertRaises(IndexError): | |
try: | |
WordMatching.getWhichLettersWereTranscribedCorrectly(word_real, mapped_letters) # , mapped_letters_indices) | |
except IndexError as e: | |
msg = 'list index out of range' | |
assert msg in str(e) | |
raise e | |
def test_get_best_mapped_words_with_empty_lists_false(self): | |
expected_words = ["?"] | |
expected_indices = [0] | |
result_words, result_indices = WordMatching.get_best_mapped_words("?", "-", use_dtw=False) | |
self.assertEqual(result_words, expected_words) | |
self.assertEqual(result_indices, expected_indices) | |
expected_words = ['b', 'i', 'n', '-'] | |
expected_indices = [np.int64(0), np.int64(1), np.int64(2), -1] | |
result_words, result_indices = WordMatching.get_best_mapped_words("bin", "bind", use_dtw=False) | |
self.assertEqual(result_words, expected_words) | |
self.assertEqual(result_indices, expected_indices) | |
def test_get_best_mapped_words_with_different_lengths_false(self): | |
result_words, result_indices = WordMatching.get_best_mapped_words("bin", "", use_dtw=False) | |
self.assertEqual(result_words, []) | |
self.assertEqual(result_indices, []) | |
def test_get_best_mapped_words_with_word_estimated_empty_real_word_not_empty_false(self): | |
result_words, result_indices = WordMatching.get_best_mapped_words("", "bin", use_dtw=False) | |
self.assertEqual(result_words, ['', '-', '-']) | |
self.assertEqual(result_indices, [-1, -1, -1]) | |
def test_get_best_mapped_words_with_word_estimated_real_word_both_empty_false(self): | |
try: | |
with self.assertRaises(IndexError): | |
try: | |
WordMatching.get_best_mapped_words("", "", use_dtw=False) | |
except IndexError as ie: | |
app_logger.error(f"raised IndexError, ie.args: {ie.args} => exception: {ie} ##") | |
msg = "index -1 is out of bounds for axis {axis} with size 0" | |
assert msg.format(axis=0) in str(ie) or msg.format(axis=1) in str(ie) | |
raise ie | |
except AssertionError: | |
# for some reason executing the test in debug mode from Visual Studio Code raises an AssertionError instead of an IndexError | |
app_logger.error("raised AssertionError instead than IndexError...") | |
with self.assertRaises(AssertionError): | |
try: | |
WordMatching.get_best_mapped_words("", "", use_dtw=False) | |
except AssertionError as ae: | |
msg = "code object dtw_low at " | |
assert msg in str(ae) | |
raise ae | |
def test_get_best_mapped_words_survived_false(self): | |
set_seed() | |
word_real = "habe" | |
for word_estimated, expected_letters, expected_indices in [ | |
("habe", ["h", "a", "b", "e"], [0, 1, 2, 3]), | |
("hobe", ["h", "-", "b", "e"], [0, -1, 2, 3]), | |
("hone", ["h", "-", "-", "e"], [0, -1, -1, 3]), | |
("honi", ["h", "-", "-", "-"], [0, -1, -1, -1]), | |
("koni", ["k", "-", "-", "-"], [0, -1, -1, -1]), | |
("kabe", ["k", "a", "b", "e"], [0, 1, 2, 3]), | |
("kane", ["k", "a", "-", "e"], [0, 1, -1, 3]), | |
]: | |
result_words, result_indices = WordMatching.get_best_mapped_words(word_estimated, word_real, use_dtw=False) | |
try: | |
self.assertEqual(result_words, expected_letters) | |
self.assertEqual(result_indices, expected_indices) | |
except AssertionError as ae: | |
app_logger.error("ae:", ae, "#", word_estimated, "#", word_real, "#", expected_letters, "#", expected_indices, "##") | |
raise ae | |
def test_get_resulting_string1(self): | |
set_seed() | |
mapped_indices = np.array([0, 1]) | |
words_estimated = ["hello", "world"] | |
words_real = ["hello", "word"] | |
expected_words = ["hello", "world"] | |
expected_indices = [0, 1] | |
result_words, result_indices = WordMatching.get_resulting_string(mapped_indices, words_estimated, words_real) | |
self.assertEqual(result_words, expected_words) | |
self.assertEqual(result_indices, expected_indices) | |
def test_get_resulting_string2(self): | |
set_seed() | |
mapped_indices = np.array([0, 1]) | |
words_estimated = ["hollo", "uorld"] | |
words_real = ["hello", "word"] | |
expected_words = ['hollo', 'uorld'] | |
expected_indices = [0, 1] | |
result_words, result_indices = WordMatching.get_resulting_string(mapped_indices, words_estimated, words_real) | |
self.assertEqual(result_words, expected_words) | |
self.assertEqual(result_indices, expected_indices) | |
mapped_indices = np.array([1, 1]) | |
expected_words = ['-', 'uorld'] | |
expected_indices = [-1, 1] | |
result_words, result_indices = WordMatching.get_resulting_string(mapped_indices, words_estimated, words_real) | |
self.assertEqual(result_words, expected_words) | |
self.assertEqual(result_indices, expected_indices) | |
mapped_indices = np.array([0, 0]) | |
expected_words = ['hollo', '-'] | |
expected_indices = [0, -1] | |
result_words, result_indices = WordMatching.get_resulting_string(mapped_indices, words_estimated, words_real) | |
self.assertEqual(result_words, expected_words) | |
self.assertEqual(result_indices, expected_indices) | |
mapped_indices = np.array([0, -1]) | |
expected_words = ["hollo", "-"] | |
expected_indices = [0, -1] | |
result_words, result_indices = WordMatching.get_resulting_string(mapped_indices, words_estimated, words_real) | |
self.assertEqual(result_words, expected_words) | |
self.assertEqual(result_indices, expected_indices) | |
mapped_indices = np.array([-1, -1]) | |
expected_words = ["-", "-"] | |
expected_indices = [-1, -1] | |
result_words, result_indices = WordMatching.get_resulting_string(mapped_indices, words_estimated, words_real) | |
self.assertEqual(result_words, expected_words) | |
self.assertEqual(result_indices, expected_indices) | |
def test_get_resulting_string_with_empty_lists(self): | |
mapped_indices = np.array([]) | |
words_estimated = [] | |
words_real = [] | |
expected_words = [] | |
expected_indices = [] | |
result_words, result_indices = WordMatching.get_resulting_string(mapped_indices, words_estimated, words_real) | |
self.assertEqual(result_words, expected_words) | |
self.assertEqual(result_indices, expected_indices) | |
if __name__ == '__main__': | |
unittest.main() | |