KASET - Kurmanji and Sorani Kurdish Speech and Transcripts
| Item Name: | KASET - Kurmanji and Sorani Kurdish Speech and Transcripts |
| Author(s): | Dana Delgado, Kevin Walker, Stephanie Strassel, David Graff, Christopher Caruso |
| LDC Catalog No.: | LDC2024S01 |
| ISLRN: | 712-302-473-619-2 |
| DOI: | https://doi.org/10.35111/576n-ye16 |
| Release Date: | January 16, 2024 |
| Member Year(s): | 2024 |
| DCMI Type(s): | Sound, Text |
| Sample Type: | CTS 16-bit pcm BN 16-bit pcm |
| Sample Rate: | CTS 8 KHz BN 16 KHz |
| Data Source(s): | broadcast news, telephone conversations |
| Project(s): | NIST SRE |
| Application(s): | language identification, speech recognition |
| Language(s): | Central Kurdish, Northern Kurdish, Kurdish |
| Language ID(s): | ckb, kmr, kur |
| License(s): |
LDC User Agreement for Non-Members |
| Online Documentation: | LDC2024S01 Documents |
| Licensing Instructions: | Subscription & Standard Members, and Non-Members |
| Citation: | Delgado, Dana, et al. KASET - Kurmanji and Sorani Kurdish Speech and Transcripts LDC2024S01. Web Download. Philadelphia: Linguistic Data Consortium, 2024. |
| Related Works: | View |
Introduction
KASET - Kurmanji and Sorani Kurdish Speech and Transcripts was developed by the Linguistic Data Consortium (LDC) and consists of approximately 147 hours of telephone conversations (289 recordings) and broadcast news (410 recordings) in two Kurdish dialects: Kurmanji Kurdish and Sorani Kurdish. Corresponding transcripts covering approximately 60 hours of the recordings are included in this release.
Kurdish is spoken primarily in Turkey, Iran, Iraq, and Syria. Sorani and Kurmanji are the two widely spoken dialects of the Kurdish language.
Data
Twenty-one native Kurdish speakers living in the continental United States each made a minimum of ten telephone calls, lasting up to ten minutes, to a family member or friend living in North America on a topic of their choice. LDC also collected multiple streaming radio and television broadcast programs (narrowband and wideband audio), many of which contained a mix of Kurmanji and Sorani Kurdish.
Telephone recordings were collected via LDC's telephone speech collection system as two-channel 8-bit μ-law, 8-KHz sample rate. Broadcast audio was captured as single channel recordings at a sample rate of 16-KHz. All audio is stored in flac-compressed format.
Native speaker auditors identified a 5-10 minute span from each broadcast recording for transcription. Full telephone recordings that passed the native speaker audit were transcribed.
The four main fields on each line on each line of the transcript files (start_offset, end_offset, speaker_label, transcript_text) are separated by tabs. Each contains a list of time-stamped segments in order according to their start_offset values, with no blank lines. The transcripts are presented as plain-text, tab-delimited files with UTF-8 character encoding.
This release includes speaker information, such as gender, year of birth, and language.
Samples
Please view these samples:
- Sorani broadcast audio (flac)
- Sorani broadcast transcript (tsv)
- Kurmanji telephone audio (flac)
- Kurmanji telephone transcript (tsv)
Updates
None at this time.