Skip to content

Samples

Arabic speech to text: a real sample transcript

What the free preview returns for a short Arabic recording (العربية), with the audio, the sources and the mistakes left in.

Updated

Arabic (العربية), 28 seconds, taken 2026-10-27. Sentences read aloud by a volunteer. Output of the free preview with the language set to Arabic, not corrected.

CC0 1.0

The transcript, as the engine returned it

  1. الأرض كوكب جميل
  2. السفر بالسفين موتع جداً
  3. هل سبق أنزلت بريطانيا؟
  4. متى تلعب التنس؟
  5. ما الذي تفعله بوقت ثراقك؟
  6. هواية يجمع اللعب القديمة
  7. ما زال لدينا الكثير من الوقت
  8. أذكر أنني أرسلت الرسالة
  9. من عقد الاجتماعه هنا
  10. يسكن عمي في الشقة
  11. ما الذي حدث هنا؟
  12. نضع العالم بين يديك

Download this transcript as SRT (written by the same exporter the app uses; the other formats are TXT, timestamped TXT, VTT, Markdown, JSON and CSV).

Against the reference sentences

Word error rate 22.4% (11 edits over 49 words), character error rate 5.6%; normalisation: Arabic (explained on the samples page).

The sentences the speaker was asked to read:

  1. الأرض كوكبٌ جميل.
  2. السفر بالسفينة ممتع جدا.
  3. هل سبق أن زرت بريطانيا ؟
  4. متى تلعب التنس ؟
  5. ما الذي تفعله بوقت فراغك ؟
  6. هوايتي جمع اللعب القديمة.
  7. ما زال لدينا الكثير من الوقت.
  8. أذكر أنني أرسلت الرسالة.
  9. انعقد الإجتماع هنا.
  10. يسكن عمي في شقة.
  11. ما الذي حدث هنا ؟
  12. نضع العالم بين يديك.
In the reference sentenceIn the transcript
بالسفينهبالسفين
ممتعموتع
ان(missing)
زرتانزلت
فراغكثراقك
هوايتيهوايه
جمعيجمع
(nothing: an extra word)من
انعقدعقد
الاجتماعالاجتماعه
شقهالشقه

Where the recording comes from

  • Source: Mozilla Common Voice 22.0, test split (CC0 1.0 public domain dedication, https://commonvoice.mozilla.org/en/datasets, https://creativecommons.org/publicdomain/zero/1.0/; copy used: https://huggingface.co/datasets/fsicoli/common_voice_22_0, whose card also says cc0-1.0).
  • Licence: CC0 1.0 (creativecommons.org/publicdomain/zero/1.0/).
  • Which recording: One volunteer speaker who labelled their accent Levantine; twelve read sentences from the Arabic test split. The same recording that plays on the Arabic home page.
  • Clip ids: common_voice_ar_22245828, common_voice_ar_22245852, common_voice_ar_22405194, common_voice_ar_22245847, common_voice_ar_22306775, common_voice_ar_22405201, common_voice_ar_22245846, common_voice_ar_22306764, common_voice_ar_22245848, common_voice_ar_22405177, common_voice_ar_22306755, common_voice_ar_22306786.

How every sample was made, step by step, is on the samples page. To try your own Arabic recording, use the Arabic transcription page.

Questions

What did the engine get wrong in this Arabic clip?
9 words were substituted, 1 were missing and 1 were extra, out of 49 reference words. For example: بالسفينه became بالسفين; ممتع became موتع; ان became (missing).
Can I use the Arabic recording?
CC0 1.0. The source is named above.

Try it with your own file

Upload a recording and read the transcript.