| <!doctype html>
|
| <html lang="en">
|
| <head>
|
| <meta charset="utf-8">
|
| <meta name="viewport" content="width=device-width,initial-scale=1">
|
| <title>Spoken Fact-Checking Datasets · TRILOGUE, Lost in Speech & MAD2</title>
|
| <meta name="description" content="Compare TRILOGUE, Lost in Speech, and MAD2: tasks, languages, dataset sizes, annotations, audio availability, and source relationships.">
|
| <meta name="theme-color" content="#142634">
|
| <link rel="icon" type="image/svg+xml" href="data:image/svg+xml,%3Csvg xmlns='http://www.w3.org/2000/svg' viewBox='0 0 32 32'%3E%3Crect width='32' height='32' rx='6' fill='%23142634'/%3E%3Cpath d='M8 9v14M16 5v22M24 11v10' stroke='%236de0cf' stroke-width='4'/%3E%3C/svg%3E">
|
| <link rel="stylesheet" href="styles.css">
|
| </head>
|
| <body>
|
| <a class="skip" href="#comparison">Skip to dataset comparison</a>
|
| <header class="site-header wrap">
|
| <a class="brand" href="#top" aria-label="Spoken Fact-Checking Datasets home"><span class="brand-mark" aria-hidden="true">III</span><span>Spoken Fact-Checking<br><strong>Dataset Guide</strong></span></a>
|
| </header>
|
| <main id="top">
|
| <section class="intro wrap" aria-labelledby="page-title">
|
|
|
| <h1 id="page-title">Find the right dataset.</h1>
|
| <p class="intro-copy">Compare three resources for fact-checking speech and dialogue.</p>
|
| </section>
|
|
|
| <section id="comparison" class="comparison wrap" aria-labelledby="compare-title">
|
| <h2 id="compare-title" class="sr-only">Compare datasets</h2>
|
| <p class="mobile-hint">Scroll the table horizontally to compare all three datasets.</p>
|
| <div class="table-scroll" tabindex="0" role="region" aria-label="Dataset comparison table; scroll horizontally on smaller screens">
|
| <table>
|
| <caption class="sr-only">Comparison of TRILOGUE, Lost in Speech, and MAD2 based on their linked dataset cards.</caption>
|
| <colgroup><col class="row-label"><col><col><col></colgroup>
|
| <thead><tr><th scope="col"></th><th scope="col" class="trilogue"><a href="https://huggingface.co/datasets/chaewanC/TRILOGUE">TRILOGUE</a><span class="subhead">Dialogue claims + evidence</span></th><th scope="col" class="lost"><a href="https://huggingface.co/datasets/maristombayeva/lost-in-speech">Lost in Speech</a><span class="subhead">Article alteration detection</span></th><th scope="col" class="mad2"><a href="https://huggingface.co/datasets/chaewanC/MAD2">MAD2</a><span class="subhead">Sentence check-worthiness</span></th></tr></thead>
|
| <tbody>
|
| <tr class="size-row"><th scope="row">Collection size</th><td><strong>11,957</strong> dialogues</td><td><strong>12,013</strong> article-derived samples<span class="cell-note">3,712 unaltered · 8,301 altered</span></td><td><strong>1,000</strong> conversations</td></tr>
|
| <tr><th scope="row">Use it to study</th><td>Which dialogue claims need checking, whether they are true, and what evidence supports them.</td><td>Whether an article-derived sample contains a factual or contextual alteration.</td><td>Which sentences in a conversation are worth fact-checking.</td></tr>
|
| <tr><th scope="row">Languages</th><td>English · Russian · Kazakh</td><td>English · Russian · Kazakh</td><td>English</td></tr>
|
| <tr><th scope="row">Audio released</th><td><b>8,036 recordings · 242.5 h</b><span class="cell-note">Russian & Kazakh; English withheld.</span></td><td><b>3,978 recordings</b><span class="cell-note">Synthetic Kazakh; English & Russian not released.</span></td><td><b>1,000 recordings · 11.04 h</b><span class="cell-note">Replacement MoonCast audio; different from the paper’s experimental recordings.</span></td></tr>
|
| </tbody>
|
| <tfoot><tr><th scope="row">Source & download</th><td><a class="dataset-link" href="https://huggingface.co/datasets/chaewanC/TRILOGUE">TRILOGUE on Hugging Face</a></td><td><a class="dataset-link" href="https://huggingface.co/datasets/maristombayeva/lost-in-speech">Lost in Speech on Hugging Face</a></td><td><a class="dataset-link" href="https://huggingface.co/datasets/chaewanC/MAD2">MAD2 on Hugging Face</a></td></tr></tfoot>
|
| </table>
|
| </div>
|
| <p class="source-note">Read the linked dataset cards for current availability, definitions, and exceptions. “Released” refers to included files, subject to each repository’s access conditions.</p>
|
| </section>
|
|
|
|
|
| <section id="relationships" class="source-section wrap" aria-labelledby="relationships-title">
|
| <h2 id="relationships-title">Source relationships & article mapping</h2>
|
| <div class="relationship-layout">
|
| <div class="linked-resources">
|
| <div class="resource-node"><span class="kicker">Generated dialogues</span><h3>TRILOGUE</h3><p>Multi-turn dialogues with check-worthiness, factuality, evidence, and paired speech.</p></div>
|
| <div class="connection"><span class="connection-line" aria-hidden="true"></span><span>Verified source mapping</span><span class="connection-line" aria-hidden="true"></span></div>
|
| <div class="resource-node"><span class="kicker">Article-derived samples</span><h3>Lost in Speech</h3><p>Unaltered articles and controlled rewrites, represented as text, synthesized speech, and ASR.</p></div>
|
| </div>
|
| <aside class="independent-resource"><span class="kicker">Separate claim source</span><h3>MAD2</h3><p>LIAR claims seed English conversations for sentence-level check-worthiness research.</p><p>MAD2 is not included in the TRILOGUE–Lost in Speech identifier mapping.</p></aside>
|
| </div>
|
| <p class="mapping-summary">The verified mapping links 11,956 dialogue inputs and 3,665 article families. Original article texts remain withheld upstream.</p>
|
| <a href="https://huggingface.co/datasets/chaewanC/TRILOGUE/blob/main/docs/source_linkage.md">Mapping methods and unresolved cases</a>
|
| </section>
|
| <section id="citations" class="citations-section wrap" aria-labelledby="citations-title">
|
| <h2 id="citations-title">Cite the datasets</h2>
|
| <article class="citation-row trilogue" aria-labelledby="cite-trilogue"><h3 id="cite-trilogue">TRILOGUE</h3><div class="citation-box"><div class="citation-toolbar"><span>BibTeX</span><button type="button" class="copy-bibtex" data-copy="bibtex-trilogue" aria-label="Copy TRILOGUE BibTeX citation" hidden>Copy BibTeX</button></div><pre tabindex="0" aria-label="TRILOGUE BibTeX citation"><code id="bibtex-trilogue">@inproceedings{chun-etal-2026-trilogue,
|
| title = {{TRILOGUE}: A Trilingual Spoken Dialogue Fact-Checking
|
| Benchmark with Evidence and Paired Audio},
|
| author = {Chun, Chaewan and Aristombayeva, Meruyert and
|
| Choi, Jiyoung and Nahar, Mahjabin and
|
| Zhang, Delvin Ce and Lee, Dongwon},
|
| booktitle = {Findings of the Association for Computational Linguistics:
|
| EMNLP 2026},
|
| year = {2026}
|
| }</code></pre><p class="copy-status" role="status" aria-live="polite"></p></div><div class="citation-links"><a href="https://huggingface.co/datasets/chaewanC/TRILOGUE#citation">Citation source on Hugging Face</a><a href="https://huggingface.co/datasets/chaewanC/TRILOGUE/blob/main/LICENSE">Research-use license</a></div><div class="citation-notes"><p>Gated research access. Source articles are linked rather than redistributed. English TTS audio is withheld; seven Kazakh human recordings have incomplete final audio.</p></div></article>
|
| <article class="citation-row lost" aria-labelledby="cite-lost"><h3 id="cite-lost">Lost in Speech</h3><div class="citation-box"><div class="citation-toolbar"><span>BibTeX</span><button type="button" class="copy-bibtex" data-copy="bibtex-lost" aria-label="Copy Lost in Speech BibTeX citation" hidden>Copy BibTeX</button></div><pre tabindex="0" aria-label="Lost in Speech BibTeX citation"><code id="bibtex-lost">@inproceedings{aristombayeva-etal-2026-lost,
|
| title = {Lost in Speech: Trilingual Spoken Hallucination Detection
|
| Across Audio and Transcripts},
|
| author = {Aristombayeva, Meruyert and Lucas, Jason S. and
|
| Chun, Chaewan and Lee, Dongwon},
|
| booktitle = {Proceedings of the Second Workshop on Speech and Audio
|
| Language Models (SALMA)},
|
| year = {2026}
|
| }</code></pre><p class="copy-status" role="status" aria-live="polite"></p></div><div class="citation-links"><a href="https://huggingface.co/datasets/maristombayeva/lost-in-speech#citation">Citation source on Hugging Face</a><a href="https://huggingface.co/datasets/maristombayeva/lost-in-speech/blob/main/LICENSE">Research-use license</a></div><div class="citation-notes"><p>Gated research access. Original texts are withheld even where ASR is available. Exclude Russian IDs 8333, 8412, and 8456 pending upstream review; their labels are not verified.</p></div></article>
|
| <article class="citation-row mad2" aria-labelledby="cite-mad2"><h3 id="cite-mad2">MAD2</h3><div class="citation-box"><div class="citation-toolbar"><span>BibTeX</span><button type="button" class="copy-bibtex" data-copy="bibtex-mad2" aria-label="Copy MAD2 BibTeX citation" hidden>Copy BibTeX</button></div><pre tabindex="0" aria-label="MAD2 BibTeX citation"><code id="bibtex-mad2">@inproceedings{chun-etal-2026-context-aware,
|
| author = {Chun, Chaewan and Zhang, Delvin Ce and Lee, Dongwon},
|
| title = {Context-Aware Multimodal Claim Verification in Spoken Dialogues},
|
| booktitle = {The 2nd Speech and Audio Language Models Workshop
|
| ({SALMA}), {EMNLP}},
|
| year = {2026}
|
| }</code></pre><p class="copy-status" role="status" aria-live="polite"></p></div><div class="citation-links"><a href="https://huggingface.co/datasets/chaewanC/MAD2#citation">Citation source on Hugging Face</a><a href="https://huggingface.co/datasets/chaewanC/MAD2/blob/main/LICENSE">Noncommercial research license</a></div><div class="citation-notes"><p>The supplied audio is a replacement corpus. Published model results and WER concern the original recordings. Current sentence labels indicate check-worthiness, not factual correctness.</p><p class="release-note">The current card says the release version remains to be finalized.</p></div></article>
|
| </section>
|
| </main>
|
| <footer class="wrap"><p>Spoken Fact-Checking Dataset Guide</p><p>Source: linked dataset cards. Information checked October 3, 2026.</p><a href="#top">Back to top</a></footer>
|
| <script src="citations.js" defer></script>
|
| </body>
|
| </html>
|
|
|
