Datasets:

Languages:
Estonian
License:

The viewer is disabled because this dataset repo requires arbitrary Python code execution. Please consider removing the loading script and relying on automated data support (you can use convert_to_parquet from the datasets library). If this is not possible, please open a discussion for direct help.

Dataset Card for "ERRnews"

Dataset Summary

ERRnews is an estonian language summarization dataset of ERR News broadcasts scraped from the ERR Archive (https://arhiiv.err.ee/err-audioarhiiv). The dataset consists of news story transcripts generated by an ASR pipeline paired with the human written summary from the archive. For leveraging larger english models the dataset includes machine translated (https://neurotolge.ee/) transcript and summary pairs.

Supported Tasks and Leaderboards

More Information Needed

Languages

Estonian

Dataset Structure

Data Instances

{'name': 'Kütuseaktsiis Balti riikides on erinev.', 'summary': 'Eestis praeguse plaani järgi järgmise aasta maini kehtiv madalam diislikütuse aktsiis ei ajenda enam tankima Lätis, kuid bensiin on seal endiselt odavam. Peaminister Kaja Kallas ja kütusemüüjad on eri meelt selles, kui suurel määral mõjutab aktsiis lõpphinda tanklais.', 'transcript': 'Eesti-Läti piiri alal on kütusehinna erinevus eriti märgatav ja ka tuntav. Õigema pildi saamiseks tuleks võrrelda ühe keti keskmist hinda, kuna tanklati võib see erineda Circle K. [...] Olulisel määral mõjutab hinda kütuste sisseost, räägib kartvski. On selge, et maailmaturuhinna põhjal tehtud ost Tallinnas erineb kütusehinnast Riias või Vilniuses või Varssavis. Kolmas mõjur ja oluline mõjur on biolisandite kasutamise erinevad nõuded riikide vahel.', 'url': 'https://arhiiv.err.ee//vaata/uudised-kutuseaktsiis-balti-riikides-on-erinev', 'meta': '\n\n\nSarja pealkiri:\nuudised\n\n\nFonoteegi number:\nRMARH-182882\n\n\nFonogrammi tootja:\n2021 ERR\n\n\nEetris:\n16.09.2021\n\n\nSalvestuskoht:\nRaadiouudised\n\n\nKestus:\n00:02:34\n\n\nEsinejad:\nKond Ragnar, Vahtrik Raimo, Kallas Kaja, Karcevskis Ojars\n\n\nKategooria:\nUudised → uudised, muu\n\n\nPüsiviide:\n\nvajuta siia\n\n\n\n', 'audio': {'path': 'recordings/12049.ogv', 'array': array([0.00000000e+00, 0.00000000e+00, 0.00000000e+00, ..., 2.44576868e-06, 6.38223427e-06, 0.00000000e+00]), 'sampling_rate': 16000}, 'recording_id': 12049}

Data Fields

name: News story headline
summary: Hand written summary.
transcript: Automatically generated transcript from the audio file with an ASR system.
url: ERR archive URL.
meta: ERR archive metadata.
en_summary: Machine translated English summary.
en_transcript: Machine translated English transcript.
audio: A dictionary containing the path to the downloaded audio file, the decoded audio array, and the sampling rate. Note that when accessing the audio column: `dataset[0]["audio"]` the audio file is automatically decoded and resampled to `dataset.features["audio"].sampling_rate`. Decoding and resampling of a large number of audio files might take a significant amount of time. Thus it is important to first query the sample index before the `"audio"` column, *i.e.* `dataset[0]["audio"]` should **always** be preferred over `dataset["audio"][0]`.
recording_id: Audio file id.

Data Splits

train validation test
10420 523 523

BibTeX entry and citation info

article{henryabstractive,
          title={Abstractive Summarization of Broadcast News Stories for {Estonian}},
          author={Henry, H{\"a}rm and Tanel, Alum{\"a}e},
          journal={Baltic J. Modern Computing},
          volume={10},
          number={3},
          pages={511-524},
          year={2022}
        }
Downloads last month
47