Elormiden's picture
Update README.md
1575ca0 verified
metadata
dataset_info:
  features:
    - name: audio
      dtype:
        audio:
          sampling_rate: 16000
    - name: text
      dtype: string
  splits:
    - name: train
      num_bytes: 3532879789.5
      num_examples: 15756
    - name: validation
      num_bytes: 396863774.75
      num_examples: 1770
    - name: test
      num_bytes: 463723737.5
      num_examples: 2068
  download_size: 4386609520
  dataset_size: 4393467301.75
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*
      - split: validation
        path: data/validation-*
      - split: test
        path: data/test-*
license: cc-by-4.0
language:
  - el
tags:
  - cypriot
  - news
  - collection

RIK Cypriot Collection Global

Dataset Description

The RIK Cypriot Collection Global is a diverse audio dataset compiled from 50+ different television programs broadcast on RIK (Cyprus Broadcasting Corporation). This comprehensive collection represents authentic Cypriot Greek speech across multiple domains, genres, and contexts. Crucially, this version of the dataset now includes high-quality text transcriptions for all audio clips. This integration transforms the collection into a robust resource primarily for Automatic Speech Recognition (ASR) model training and evaluation, as well as continued linguistic and dialectal studies.

Dataset Details

  • Language: Greek (Cypriot dialect)
  • Source: RIK (Cyprus Broadcasting Corporation)
  • Programs: 50+ different television shows
  • Sampling Rate: 16,000 Hz
  • Total Examples: 19,594 audio-text pairs
  • Dataset Size: ~4.1 GB
  • Coverage: Multi-domain, diverse content

Data Fields

  • audio: Audio recordings from various TV programs (16kHz sampling rate)

Data Splits

  • Training set: 15,756 samples
  • Validation set: 1,770 samples
  • Test set: 2,068 samples

Total: 19,594 audio-text pairs

Licensing and Usage

All audio recordings in this dataset are the property of the Cyprus Broadcasting Corporation (RIK). This dataset is provided strictly for research and educational purposes. The author does not claim ownership of the original audio content. Transcriptions, segmentations, and dataset structuring were created by the author.

For any commercial use of the original audio material, please contact RIK directly.