Sign inGet started
Dataset · Hugging Face

FatimahEmadEldin/arabic-emotion-voice-vs-words

Voice vs Words · الصوت أم الكلمات Does a compact audio-language model hear how something is said in Saudi Arabic, or only read what is said? Link Evaluation set (288 clips, labels, audio) data/ · dataset viewer Model outputs, tables, figures results/ Blind native-listener validation results/human/ Annotation dashboard (listen, rate, export) huggingface.co/spaces/FatimahEmadEldin/voice-vs-words-annotation Code and how to run it scripts/ ·… See the full description on the dataset page: https://huggingface.co/datasets/FatimahEmadEldin/arabic-emotion-voice-vs-words.

Author / organizationFatimahEmadEldin
License (as declared)cc-by-4.0
Downloads0 (as reported by Hugging Face)
Likes0
Created22 October 2026
Last updated at source22 October 2026
speech-emotion-recognitionparalinguisticsarabicsaudi-dialectaudio-language-modelsevaluationblind-spotqwen2.5-omni
Embed this entry
<iframe src="https://silytics.si/embed/item/638b736f-72a5-4087-9415-28d65256d09e" width="420" height="190" style="border:0" loading="lazy" title="Silytics entry"></iframe>

Free, no tracking. More about feeds and embeds.

Recent changes

  • 22 Oct 2026 Description: Voice vs Words · الصوت أم الكلمات Does a compact audio-language model hear how something is said in Saudi Arabic, or only read what is said? Overview Qwen2.5-Omni-3B recognises happy, angry and sad speech in Saudi Arabic almost perfectly (9 → Voice vs Words · الصوت أم الكلمات Does a compact audio-language model hear how something is said in Saudi Arabic, or only read what is said? Link Evaluation set (288 clips, labels, audio) data/ · dataset viewer Model outputs, tables, figure
  • 22 Oct 2026 License: none → cc-by-4.0
  • 22 Oct 2026 Description: none → Voice vs Words · الصوت أم الكلمات Does a compact audio-language model hear how something is said in Saudi Arabic, or only read what is said? Overview Qwen2.5-Omni-3B recognises happy, angry and sad speech in Saudi Arabic almost perfectly (9
  • 22 Oct 2026 Topics added task_categories:audio-classification, speech-emotion-recognition, paralinguistics, arabic, saudi-dialect, audio-language-models, evaluation, blind-spot, qwen2.5-omni

Changes detected by Silytics between refreshes of the source.

More from FatimahEmadEldin

Silytics shows public metadata from Hugging Face and links to the original. Information may be incomplete or out of date. Last refreshed 22 October 2026. Licenses and terms of the original apply to the work itself. Not financial, legal or professional advice.