retkowski/ytseg

From Text Segmentation to Smart Chaptering: A Novel Benchmark for Structuring Video Transcriptions

We present YTSeg, a topically and structurally diverse benchmark for the text segmentation task based on YouTube transcriptions. The dataset comprises 19,299 videos from 393 channels, amounting to 6,533 content hours. The topics are wide-ranging, covering domains such as science, lifestyle, politics, health, economy, and technology. The videos are from various types of content formats, such as podcasts, lectures, news, corporate events & promotional content, and, more broadly, videos from individual content creators. We refer to the paper (acl | arXiv) for further information. We provide both text and audio data as well as a download script for the video data.

Data Overview

We offer three dataset subsets:

• Text — For text-based segmentation and chaptering approaches using transcripts.
• Audio — For audio-based chaptering approaches with embedded audio.
• Titles — For chapter title generation given segment text (relevant for two-stage approaches).

YTSeg (Text)

Each video is represented as a JSON object. The fields are organized into three categories: Transcripts, Target Representations, and Metadata.

Transcripts

We provide three transcript variants for each video: the original reference transcript and two ASR-generated transcripts using Whisper models.

Field	Description
text_ref	Reference transcript as a flat list of sentences.
text_wt	Whisper-tiny ASR transcript as a flat list of sentences.
text_wl	Whisper-large ASR transcript as a flat list of sentences.

Target Representations

Multiple target formats are provided for different modeling approaches.

Field	Description
target_binary_ref	Binary segmentation labels for reference transcript (e.g., |=000100000010).
target_binary_wt	Binary segmentation labels for Whisper-tiny transcript.
target_binary_wl	Binary segmentation labels for Whisper-large transcript.
target_text_ref	Structured transcript with chapter markers (e.g., [CSTART] Title [CEND] text...).
target_text_ts_ref	Structured transcript with timestamped chapter markers (e.g., [CSTART] 00:01:23 - Title [CEND] text...).
target_ts	Timestamped chapter markers only (e.g., [CSTART] 00:01:23 - Title [CEND]\n...).

Metadata

Field	Description
audio_path	Path to the .mp3 file of the video.
chapter_titles	A list of chapter titles corresponding to each segment.
channel_id	The YouTube channel ID which this video belongs to.
video_id	The YouTube video ID.
speaker_category	Speaker classification: single, single_weak, or multiple.
dominant_speaker_proportion	Proportion of speech from the dominant speaker (0.0-1.0).
num_speakers	Number of detected speakers in the video.
duration	Video duration in seconds.

Partition Statistics

Partition	# Examples
Training	16,404 (85%)
Validation	1,447 (7.5%)
Testing	1,448 (7.5%)
Total	19,299

YTSeg (Audio)

The audio config provides the complete dataset with embedded audio files. Each video is represented with the same fields as the text config, plus an audio field containing the preprocessed audio data.

Audio

Field	Description
audio	Audio data preprocessed into .mp3 format with a standardized sample rate of 16,000 Hz and a single channel (mono).

All other fields (transcripts, target representations, and metadata) are identical to the Text config described above.

Partition Statistics

Partition	# Examples	Size (GB)
Training	16,404 (85%)	~57.9 GB
Validation	1,447 (7.5%)	~5.1 GB
Testing	1,448 (7.5%)	~5.3 GB
Total	19,299	~68.3 GB

YTSeg (Titles)

Each chapter of a video is represented as a JSON object with the following fields:

Field	Description
text_section_ref	The complete chapter/section text.
text_section_prev_titles_ref	The complete chapter/section text with previous section titles prepended.
target_title	The target chapter title.
channel_id	The YouTube channel ID which this chapter's video belongs to.
video_id	The YouTube video ID which this chapter belongs to.
chapter_idx	The index and placement of the chapter in the video (e.g., the first chapter has index 0).

Partition	# Examples
Training	146,907 (84.8%)
Validation	13,206 (7.6%)
Testing	13,082 (7.6%)
Total	173,195

Video Data

A download script for the video and audio data is provided.

python download_videos.py

In the script, you can further specify a target folder (default is ./video) and target formats in a priority list.

Loading Data

The dataset can be loaded directly using the HuggingFace datasets library:

from datasets import load_dataset

# Load the audio config (with embedded audio)
dataset = load_dataset("retkowski/ytseg", "audio", split="test")

# Load the text config (text-only)
dataset = load_dataset("retkowski/ytseg", "text", split="test")

# Load the titles config
dataset = load_dataset("retkowski/ytseg", "titles", split="test")

Note on Binary Labels: The binary segmentation labels (e.g., target_binary_ref) are prefixed with |= to force the field to be stored as a string, preventing leading zeros from being lost during processing. For actual usage, strip the |= prefix:

binary_labels = dataset['target_binary_ref'].lstrip('|=')

Citing

We kindly request you to cite our corresponding EACL 2024 paper if you use our dataset.

@inproceedings{retkowski-waibel-2024-text,
    title = "From Text Segmentation to Smart Chaptering: A Novel Benchmark for Structuring Video Transcriptions",
    author = "Retkowski, Fabian  and Waibel, Alexander",
    editor = "Graham, Yvette  and Purver, Matthew",
    booktitle = "Proceedings of the 18th Conference of the European Chapter of the Association for Computational Linguistics (Volume 1: Long Papers)",
    month = mar,
    year = "2024",
    address = "St. Julian{'}s, Malta",
    publisher = "Association for Computational Linguistics",
    url = "https://aclanthology.org/2024.eacl-long.25",
    pages = "406--419",
    abstract = "Text segmentation is a fundamental task in natural language processing, where documents are split into contiguous sections. However, prior research in this area has been constrained by limited datasets, which are either small in scale, synthesized, or only contain well-structured documents. In this paper, we address these limitations by introducing a novel benchmark YTSeg focusing on spoken content that is inherently more unstructured and both topically and structurally diverse. As part of this work, we introduce an efficient hierarchical segmentation model MiniSeg, that outperforms state-of-the-art baselines. Lastly, we expand the notion of text segmentation to a more practical {``}smart chaptering{''} task that involves the segmentation of unstructured content, the generation of meaningful segment titles, and a potential real-time application of the models.",
}

Changelog

• 20.01.2025 -- Major data and format update:
  • Added ASR transcripts (Whisper-tiny and Whisper-large), structured transcript targets with timestamps, and metadata for finer-grained analysis (speaker category, dominant speaker proportion, number of speakers, duration)
  • Added audio config with HuggingFace Audio feature for seamless loading with embedded audio
  • Updated to use HuggingFace datasets library for data loading (replacing local pandas scripts and use proper HF configs)
  • Updated YTSeg[Titles] field names for clarity
• 25.07.2024 -- Added complete list of chapter titles to YTSeg (YTSeg[Titles] is a filtered subset)
• 09.04.2024 -- Added audio data
• 27.02.2024 -- Initial release

License

The dataset is available under the Creative Commons Attribution-NonCommercial-ShareAlike (CC BY-NC-SA) 4.0 license. We note that we do not own the copyright of the videos and as such opted to release the dataset with a non-commercial license, with the intended use to be in research and education.