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 Nota Lyd- og tekstdata
Dataset Summary
This data was created by the public institution Nota (https://nota.dk/), which is part of the Danish Ministry of Culture. Nota has a library audiobooks and audiomagazines for people with reading or sight disabilities. Nota also produces a number of audiobooks and audiomagazines themselves.
The dataset consists of .wav and .txt files from Nota's audiomagazines "Inspiration" and "Radio/TV".
The dataset has been published as a part of the initiative sprogteknologi.dk, within the Danish Agency for Digital Government (www.digst.dk).
336 GB available data, containing voice recordings and accompanying transcripts.
Each publication has been segmented into bits of 2 - 50 seconds .wav files with an accompanying transcription
Supported Tasks and Leaderboards
[Needs More Information]
Languages
Danish
Dataset Structure
Data Instances
A typical data point comprises the path to the audio file, called path and its sentence.
{'path': '<path_to_clip>.wav', 'sentence': 'Dette er et eksempel', 'audio': {'path': <path_to_clip>.wav', 'array': array([-0.00048828, -0.00018311, -0.00137329, ..., 0.00079346, 0.00091553, 0.00085449], dtype=float32), 'sampling_rate': 44100}
Data Fields
path: The path to the audio file
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].
sentence: The sentence that was read by the speaker
Data Splits
The material has for now only a train split. As this is very early stage of the dataset, splits might be introduced at a later stage.
Dataset Creation
Disclaimer
There might be smaller discrepancies between the .wav and .txt files. Therefore, there might be issues in the alignment of timestamps, text and sound files.
There are no strict rules as to how readers read aloud non-letter characters (i.e. numbers, €, $, !, ?). These symbols can be read differently throughout the dataset.
Curation Rationale
[Needs More Information]
Source Data
Initial Data Collection and Normalization
[Needs More Information]
Who are the source language producers?
[Needs More Information]
Annotations
Annotation process
[Needs More Information]
Who are the annotators?
[Needs More Information]
Personal and Sensitive Information
The dataset is made public and free to use. Recorded individuals has by written contract accepted and agreed to the publication of their recordings. Other names appearing in the dataset are already publically known individuals (i.e. TV or Radio host names). Their names are not to be treated as sensitive or personal data in the context of this dataset.
Considerations for Using the Data
Social Impact of Dataset
[More Information Needed]
Discussion of Biases
[More Information Needed]
Other Known Limitations
[More Information Needed]
Additional Information
Dataset Curators
Contact info@sprogteknologi.dk if you have questions regarding use of data. They gladly receive inputs and ideas on how to distribute the data.
Licensing Information
- Downloads last month
- 46