chop videos in chunks

yttemporal180m.py

@@ -20,15 +20,16 @@ _URL_BASE = "https://rowanzellers.com/merlot/#data"
 
 url_numbers = ["00" + str(i) if i < 10 else "0" + str(i) for i in range(100)]
 _DL_URLS = [
-    f"https://storage.googleapis.com/merlot/yttemporal180m/yttemporal180m_{num}of100.jsonl.gz" for num in url_numbers]
+    f"https://storage.googleapis.com/merlot/yttemporal180m/yttemporal180m_{num}of100.jsonl.gz"
+    for num in url_numbers
+]
 
 
 def json_serializer(o):
     if isinstance(o, datetime):
         return str(o)
 
-    raise TypeError(
-        f"Object of type {o.__class__.__name__} is not JSON serializable")
+    raise TypeError(f"Object of type {o.__class__.__name__} is not JSON serializable")
 
 
 class yttemporal180mConfig(datasets.BuilderConfig):
@@ -36,7 +37,8 @@ class yttemporal180mConfig(datasets.BuilderConfig):
 
     def __init__(self, **kwargs):
         super(yttemporal180mConfig, self).__init__(
-            version=datasets.Version("2.1.0", ""), **kwargs)
+            version=datasets.Version("2.1.0", ""), **kwargs
+        )
 
 
 class yttemporal180m(datasets.GeneratorBasedBuilder):
@@ -44,7 +46,8 @@ class yttemporal180m(datasets.GeneratorBasedBuilder):
     DEFAULT_CONFIG_NAME = "default"
     BUILDER_CONFIGS = [
         yttemporal180mConfig(
-            name="default", description="Default full yttemporal180m dataset"),
+            name="default", description="Default full yttemporal180m dataset"
+        ),
     ]
 
     def _info(self):
@@ -54,8 +57,9 @@ class yttemporal180m(datasets.GeneratorBasedBuilder):
                 {
                     "video_id": datasets.Value("string"),
                     "video_url": datasets.Value("string"),
-                    "asr": datasets.Value("string"),
-                    "title": datasets.Value("string"),
+                    "caption": datasets.Value("string"),
+                    "timestamp_start": datasets.Value("float32"),
+                    "timestamp_stop": datasets.Value("float32"),
                     "meta": datasets.Value("string"),
                 }
             ),
@@ -65,15 +69,12 @@ class yttemporal180m(datasets.GeneratorBasedBuilder):
         )
 
     def _split_generators(self, dl_manager):
-        archive_paths = [dl_manager.download_and_extract(
-            url) for url in _DL_URLS]
+        archive_paths = [dl_manager.download_and_extract(url) for url in _DL_URLS]
 
         train_split = [
             datasets.SplitGenerator(
                 name=datasets.Split.TRAIN,
-                gen_kwargs={
-                    "jsonl_files": archive_paths
-                },
+                gen_kwargs={"jsonl_files": archive_paths},
             )
         ]
 
@@ -88,26 +89,47 @@ class yttemporal180m(datasets.GeneratorBasedBuilder):
                 for json_str in json_list:
                     infos = json.loads(json_str)
 
-                    id = infos['info']['display_id']
+                    id = infos["info"]["display_id"]
                     url = "https://www.youtube.com/watch?v=" + id
-                    asr = ""
-                    for example in infos['denoised']:
-                        asr += example["cleanasr"]
-
-                    metadata_dict = {
-                        "asr_info": infos["denoised"],
-                        "info": infos["info"],
-                        "subtitles": infos["subtitles"],
-                    }
-                    yield idx, {
-                        "video_id": id,
-                        "video_url": url,
-                        "asr": asr,
-                        "title": infos['info']['title'],
-                        "meta": json.dumps(
-                            metadata_dict,
-                            default=json_serializer,
-                            indent=2
+
+                    # Divide video by segments of 15 sec
+                    max_sec_per_segment = 15
+                    num_chunks = (
+                        int(
+                            divmod(infos["subtitles"][-1]["time"], max_sec_per_segment)[
+                                0
+                            ]
                         )
-                    }
-                    idx += 1
+                        + 1
+                    )
+                    time_chunks = [
+                        i * max_sec_per_segment for i in range(num_chunks + 1)
+                    ]
+                    time_chunk_idx = 0
+                    caption = ""
+                    for el in infos["subtitles"]:
+                        if el["time"] <= time_chunks[time_chunk_idx + 1]:
+                            caption += el["word"] + " "
+                        else:
+                            timestamp_start = float(time_chunks[time_chunk_idx])
+                            timestamp_stop = float(time_chunks[time_chunk_idx + 1])
+                            time_chunk_idx += 1
+
+                            metadata_dict = {
+                                "asr_info": infos["denoised"],
+                                "info": infos["info"],
+                                "subtitles": infos["subtitles"],
+                                "title": infos["info"]["title"],
+                            }
+                            yield idx, {
+                                "video_id": id,
+                                "video_url": url,
+                                "caption": caption,
+                                "timestamp_start": timestamp_start,
+                                "timestamp_stop": timestamp_stop,
+                                "meta": json.dumps(
+                                    metadata_dict, default=json_serializer, indent=2
+                                ),
+                            }
+                            idx += 1
+                            caption = ""