Common Voice Scripted Speech 25.0 - Dhatki

mki — Dhatki (mki)

This datasheet is for cv-corpus-25.0-2026-03-09 of the Mozilla Common Voice Scripted Speech dataset for Dhatki [mki - mki]. The dataset contains 11325 clips representing 10.04 hours of recorded speech (9.92 hours validated) from 12 speakers, recorded from a text corpus of 2,055 sentences.

Language

Dhatki is a regional Indo-Aryan language, mainly spoken in Sindh (Pakistan) and Rajasthan (India). It is very close to Sindhi and Rajasthani, written in different scripts depending on the region, and mostly used in daily conversation, folk songs, and local stories.

Demographic information

The dataset includes the following self-declared age and gender distributions. A coverage summary is shown below each table.

Gender

Self-declared gender information. The table shows clip and speaker counts with percentages. Speakers who did not declare a gender are listed as Unspecified. A dash (-) indicates zero.

Gender declared: 2,048 of 11,325 clips (18.1%), 1 of 12 speakers (8.3%)

Age

Self-declared age information. The table shows clip and speaker counts with percentages. Speakers who did not declare an age are listed as Unspecified. A dash (-) indicates zero.

Age declared: 11,298 of 11,325 clips (99.8%), 8 of 12 speakers (66.7%)

Data splits for modelling

Clip buckets

Training splits

Training split coverage: 2,031 of 11,184 validated clips (18.2%)

The dataset contains 11184 validated, 135 invalidated, and 6 unresolved clips. The average clip duration is 3.194 seconds.

Text corpus

The Dhatki corpus used in this project consists of original sentences written by the researcher. Unlike other collections that rely on religious texts or folk stories, this corpus is self-constructed (Rishi Parkash) to represent natural and everyday usage of Dhatki.

Validated sentences: 2,055

The corpus contains 2,055 sentences: 2,055 validated and 0 unvalidated (0 pending review, 0 rejected), with 0 reported for review.

Writing system

The corpus is written in the Sindhi script. While Dhatki can be written in either Perso-Arabic (Sindhi-based) or Devanagari, this dataset follows the Sindhi convention, reflecting the variety spoken in Pakistan (Sindh).

Symbol table

My corpus is written Sindhi, Sindhi script has 52 letters. ا ب ٻ پ ڀ ت ٿ ٽ ٺ ث ج ڄ جھ چ ڇ ح خ د ڌ ڊڍ ڏ ڍ ذر ر ڙ ز ژ س ش ص ض ط ظ ع غ ف ڦ ق ڪ گ ڳ ڱ ل ڻ ن ڃ م ء و ؤ ۏ ه ء ي ءَ ءِ ءُ

Sample

There follows a randomly selected sample of five sentences from the corpus.

1. هون نئون ڪوٽ جائون تو.

2. اي شاعري لکي ٿو.

3. گيس ماهنگي هي.

4. منا واپار ڪرڻ آئي ٿو.

5. اٿي سڪون واڙو ڪم هي.

Sources

The corpus is entirely based on original sentences written by the Rishi Parkash in Dhatki (Sindhi script), reflecting natural language use. No external written sources were used.

Text domains

General, Service and Retail, Healthcare, Nature and Environment, Language Fundamentals (e.g. Digits, Letters, Money)

Processing

The corpus was created by composing original sentences in Dhatki, typing them in Sindhi script, and then editing for accuracy and consistency. Sentences were cleaned, standardized, and organized into a digital format for use in analysis.

Fields

Clips

Each row of a tsv file represents a single audio clip, and contains the following information:

• client_id - hashed UUID of a given user

• path - relative path of the audio file

• text - supposed transcription of the audio

• up_votes - number of people who said audio matches the text

• down_votes - number of people who said audio does not match text

• age - age of the speaker1

• gender - gender of the speaker1

• accents - accents of the speaker1

• variant - variant of the language1

• segment - if sentence belongs to a custom dataset segment, it will be listed here

• prompt_upvotes - number of upvotes the sentence prompt received

• prompt_reports - number of reports the sentence prompt received

• is_edited - whether the clip's transcription has been edited

validated_sentences.tsv

The validated_sentences.tsv file contains one row per validated sentence in the text corpus:

• sentence_id - unique identifier for the sentence

• sentence - the sentence text

• variant - the variant of the language

• sentence_domain - the domain(s) the sentence belongs to

• source - the source the sentence was collected from

• is_used - whether the sentence is still in circulation for recording

• clips_count - number of clips recorded for this sentence

unvalidated_sentences.tsv

The unvalidated_sentences.tsv file contains one row per unvalidated sentence in the text corpus:

• sentence_id - unique identifier for the sentence

• sentence - the sentence text

• variant - the variant of the language

• sentence_domain - the domain(s) the sentence belongs to

• source - the source the sentence was collected from

• up_votes - number of upvotes the sentence received

• down_votes - number of downvotes the sentence received

• status - current status of the sentence (pending or rejected)

Get involved

Community links

• Common Voice translators on Pontoon

• Common Voice Communities

Discussions

• Common Voice on Matrix

• Common Voice on Discourse

• Common Voice on Discord

• Common Voice on Telegram

Contribute

• Speak

• Write

• Listen

• Review

Acknowledgements

Datasheet authors

Common Voice Community

Funding

This dataset was partially funded by the Open Multilingual Speech Fund managed by Mozilla Common Voice.

Licence

This dataset is released under the Creative Commons Zero (CC-0) licence. By downloading this data you agree to not determine the identity of speakers in the dataset.

Footnotes

1. For a full list of age, gender, and accent options, see the demographics spec. These will only be reported if the speaker opted in to provide that information. ↩ ↩2 ↩3 ↩4