From d372d354cfde005dfcf85f115940f445c44cffc8 Mon Sep 17 00:00:00 2001 From: BenAAndrew Date: Mon, 9 Aug 2021 16:40:58 +0100 Subject: [PATCH] Rename aligned to text --- dataset/clip_generator.py | 6 +++--- tests/test_dataset.py | 2 +- 2 files changed, 4 insertions(+), 4 deletions(-) diff --git a/dataset/clip_generator.py b/dataset/clip_generator.py index ca5d137..140a4bf 100644 --- a/dataset/clip_generator.py +++ b/dataset/clip_generator.py @@ -67,7 +67,7 @@ def _generate_clips_from_textfile( ): fragment_matched = clean_text[fragment["match-start"] : fragment["match-end"]] if fragment_matched: - fragment["aligned"] = fragment_matched + fragment["text"] = fragment_matched clip_lengths.append((fragment["end"] - fragment["start"]) // 1000) result_fragments.append(fragment) @@ -113,7 +113,7 @@ def _generate_clips_from_subtitles( if score >= min_confidence: result_fragments.append({ "name": filename, - "aligned": text, + "text": text, "score": score, "duration": duration }) @@ -225,7 +225,7 @@ def clip_generator( # Produce metadata file with open(label_path, "w", encoding=CHARACTER_ENCODING) as f: for fragment in clips: - f.write(f"{fragment['name']}|{fragment['aligned']}\n") + f.write(f"{fragment['name']}|{fragment['text']}\n") logging.info("Generated clips") return clip_lengths diff --git a/tests/test_dataset.py b/tests/test_dataset.py index cfeb5e2..aeb33eb 100644 --- a/tests/test_dataset.py +++ b/tests/test_dataset.py @@ -63,7 +63,7 @@ def test_create_dataset(): with open(forced_alignment_path, "r") as forced_alignment_file: data = json.load(forced_alignment_file) for segment in data: - assert {"start", "end", "name", "score", "aligned"}.issubset( + assert {"start", "end", "name", "score", "text"}.issubset( segment.keys() ), "Alignment JSON missing required keys" assert segment["score"] >= min_confidence, "SWS score less than min confidence"