Rename utterance to transcript in datasets (#1841)

test/torchaudio_unittest/datasets/cmuarctic_test.py

@@ -18,7 +18,7 @@ def get_mock_dataset(root_dir):
     """
     mocked_data = []
     sample_rate = 16000
-    utterance = "This is a test utterance."
+    transcript = "This is a test transcript."
 
     base_dir = os.path.join(root_dir, "ARCTIC", "cmu_us_aew_arctic")
     txt_dir = os.path.join(base_dir, "etc")
@@ -44,11 +44,11 @@ def get_mock_dataset(root_dir):
                 sample = (
                     normalize_wav(data),
                     sample_rate,
-                    utterance,
+                    transcript,
                     utterance_id.split("_")[1],
                 )
                 mocked_data.append(sample)
-                txt.write(f'( {utterance_id} "{utterance}" )\n')
+                txt.write(f'( {utterance_id} "{transcript}" )\n')
                 seed += 1
     return mocked_data
 
@@ -66,10 +66,10 @@ class TestCMUARCTIC(TempDirMixin, TorchaudioTestCase):
 
     def _test_cmuarctic(self, dataset):
         n_ite = 0
-        for i, (waveform, sample_rate, utterance, utterance_id) in enumerate(dataset):
+        for i, (waveform, sample_rate, transcript, utterance_id) in enumerate(dataset):
             expected_sample = self.samples[i]
             assert sample_rate == expected_sample[1]
-            assert utterance == expected_sample[2]
+            assert transcript == expected_sample[2]
             assert utterance_id == expected_sample[3]
             self.assertEqual(expected_sample[0], waveform, atol=5e-5, rtol=1e-8)
             n_ite += 1

test/torchaudio_unittest/datasets/librispeech_test.py

@@ -11,7 +11,7 @@ from torchaudio_unittest.common_utils import (
 
 from torchaudio.datasets import librispeech
 
-# Used to generate a unique utterance for each dummy audio file
+# Used to generate a unique transcript for each dummy audio file
 _NUMBERS = [
     'ZERO',
     'ONE',
@@ -51,11 +51,11 @@ def get_mock_dataset(root_dir):
                 filename = f'{speaker_id}-{chapter_id}-{utterance_id:04d}.wav'
                 path = os.path.join(chapter_path, filename)
 
-                utterance = ' '.join(
+                transcript = ' '.join(
                     [_NUMBERS[x] for x in [speaker_id, chapter_id, utterance_id]]
                 )
                 trans_content.append(
-                    f'{speaker_id}-{chapter_id}-{utterance_id:04d} {utterance}'
+                    f'{speaker_id}-{chapter_id}-{utterance_id:04d} {transcript}'
                 )
 
                 data = get_whitenoise(
@@ -69,7 +69,7 @@ def get_mock_dataset(root_dir):
                 sample = (
                     normalize_wav(data),
                     sample_rate,
-                    utterance,
+                    transcript,
                     speaker_id,
                     chapter_id,
                     utterance_id
@@ -104,11 +104,11 @@ class TestLibriSpeech(TempDirMixin, TorchaudioTestCase):
     def _test_librispeech(self, dataset):
         num_samples = 0
         for i, (
-                data, sample_rate, utterance, speaker_id, chapter_id, utterance_id
+                data, sample_rate, transcript, speaker_id, chapter_id, utterance_id
         ) in enumerate(dataset):
             self.assertEqual(data, self.samples[i][0], atol=5e-5, rtol=1e-8)
             assert sample_rate == self.samples[i][1]
-            assert utterance == self.samples[i][2]
+            assert transcript == self.samples[i][2]
             assert speaker_id == self.samples[i][3]
             assert chapter_id == self.samples[i][4]
             assert utterance_id == self.samples[i][5]

test/torchaudio_unittest/datasets/vctk_test.py

@@ -11,8 +11,8 @@ from torchaudio_unittest.common_utils import (
     normalize_wav,
 )
 
-# Used to generate a unique utterance for each dummy audio file
-_UTTERANCE = [
+# Used to generate a unique transcript for each dummy audio file
+_TRANSCRIPT = [
     'Please call Stella',
     'Ask her to bring these things',
     'with her from the store',
@@ -59,14 +59,14 @@ def get_mock_dataset(root_dir):
             save_wav(audio_file_path, data, sample_rate)
 
             txt_file_path = os.path.join(file_dir, filename[:-5] + '.txt')
-            utterance = _UTTERANCE[utterance_id - 1]
+            transcript = _TRANSCRIPT[utterance_id - 1]
             with open(txt_file_path, 'w') as f:
-                f.write(utterance)
+                f.write(transcript)
 
             sample = (
                 normalize_wav(data),
                 sample_rate,
-                utterance,
+                transcript,
                 speaker_id,
                 utterance_id
             )
@@ -88,10 +88,10 @@ class TestVCTK(TempDirMixin, TorchaudioTestCase):
 
     def _test_vctk(self, dataset):
         num_samples = 0
-        for i, (data, sample_rate, utterance, speaker_id, utterance_id) in enumerate(dataset):
+        for i, (data, sample_rate, transcript, speaker_id, utterance_id) in enumerate(dataset):
             self.assertEqual(data, self.samples[i][0], atol=5e-5, rtol=1e-8)
             assert sample_rate == self.samples[i][1]
-            assert utterance == self.samples[i][2]
+            assert transcript == self.samples[i][2]
             assert speaker_id == self.samples[i][3]
             assert int(utterance_id) == self.samples[i][4]
             num_samples += 1

torchaudio/datasets/cmuarctic.py

@@ -58,10 +58,10 @@ def load_cmuarctic_item(line: str,
                         folder_audio: str,
                         ext_audio: str) -> Tuple[Tensor, int, str, str]:
 
-    utterance_id, utterance = line[0].strip().split(" ", 2)[1:]
+    utterance_id, transcript = line[0].strip().split(" ", 2)[1:]
 
-    # Remove space, double quote, and single parenthesis from utterance
-    utterance = utterance[1:-3]
+    # Remove space, double quote, and single parenthesis from transcript
+    transcript = transcript[1:-3]
 
     file_audio = os.path.join(path, folder_audio, utterance_id + ext_audio)
 
@@ -71,7 +71,7 @@ def load_cmuarctic_item(line: str,
     return (
         waveform,
         sample_rate,
-        utterance,
+        transcript,
         utterance_id.split("_")[1]
     )
 
@@ -164,7 +164,7 @@ class CMUARCTIC(Dataset):
             n (int): The index of the sample to be loaded
 
         Returns:
-            tuple: ``(waveform, sample_rate, utterance, utterance_id)``
+            tuple: ``(waveform, sample_rate, transcript, utterance_id)``
         """
         line = self._walker[n]
         return load_cmuarctic_item(line, self._path, self._folder_audio, self._ext_audio)

torchaudio/datasets/librispeech.py

@@ -49,7 +49,7 @@ def load_librispeech_item(fileid: str,
     # Load text
     with open(file_text) as ft:
         for line in ft:
-            fileid_text, utterance = line.strip().split(" ", 1)
+            fileid_text, transcript = line.strip().split(" ", 1)
             if fileid_audio == fileid_text:
                 break
         else:
@@ -59,7 +59,7 @@ def load_librispeech_item(fileid: str,
     return (
         waveform,
         sample_rate,
-        utterance,
+        transcript,
         int(speaker_id),
         int(chapter_id),
         int(utterance_id),
@@ -133,7 +133,7 @@ class LIBRISPEECH(Dataset):
             n (int): The index of the sample to be loaded
 
         Returns:
-            tuple: ``(waveform, sample_rate, utterance, speaker_id, chapter_id, utterance_id)``
+            tuple: ``(waveform, sample_rate, transcript, speaker_id, chapter_id, utterance_id)``
         """
         fileid = self._walker[n]
         return load_librispeech_item(fileid, self._path, self._ext_audio, self._ext_txt)

torchaudio/datasets/vctk.py

@@ -110,7 +110,7 @@ class VCTK_092(Dataset):
         return torchaudio.load(file_path)
 
     def _load_sample(self, speaker_id: str, utterance_id: str, mic_id: str) -> SampleType:
-        utterance_path = os.path.join(
+        transcript_path = os.path.join(
             self._txt_dir, speaker_id, f"{speaker_id}_{utterance_id}.txt"
         )
         audio_path = os.path.join(
@@ -120,12 +120,12 @@ class VCTK_092(Dataset):
         )
 
         # Reading text
-        utterance = self._load_text(utterance_path)
+        transcript = self._load_text(transcript_path)
 
         # Reading FLAC
         waveform, sample_rate = self._load_audio(audio_path)
 
-        return (waveform, sample_rate, utterance, speaker_id, utterance_id)
+        return (waveform, sample_rate, transcript, speaker_id, utterance_id)
 
     def __getitem__(self, n: int) -> SampleType:
         """Load the n-th sample from the dataset.
@@ -134,7 +134,7 @@ class VCTK_092(Dataset):
             n (int): The index of the sample to be loaded
 
         Returns:
-            tuple: ``(waveform, sample_rate, utterance, speaker_id, utterance_id)``
+            tuple: ``(waveform, sample_rate, transcript, speaker_id, utterance_id)``
         """
         speaker_id, utterance_id = self._sample_ids[n]
         return self._load_sample(speaker_id, utterance_id, self._mic_id)