| --- |
| language: |
| - da |
| license: openrail |
| size_categories: |
| - 100K<n<1M |
| task_categories: |
| - automatic-speech-recognition |
| - audio-classification |
| pretty_name: CoRal-v2 |
| dataset_info: |
| - config_name: conversation |
| features: |
| - name: id_conversation |
| dtype: string |
| - name: location |
| dtype: string |
| - name: location_roomdim |
| dtype: string |
| - name: noise_level |
| dtype: string |
| - name: noise_type |
| dtype: string |
| - name: id_speaker |
| dtype: string |
| - name: age |
| dtype: int64 |
| - name: gender |
| dtype: string |
| - name: dialect |
| dtype: string |
| - name: country_birth |
| dtype: string |
| - name: education |
| dtype: string |
| - name: occupation |
| dtype: string |
| - name: id_segment |
| dtype: int64 |
| - name: text |
| dtype: string |
| - name: audio |
| dtype: audio |
| splits: |
| - name: train |
| num_bytes: 15590409443.832 |
| num_examples: 43004 |
| - name: val |
| num_bytes: 369507908.0 |
| num_examples: 939 |
| - name: test |
| num_bytes: 446358150.838 |
| num_examples: 1259 |
| download_size: 10141070853 |
| dataset_size: 16406275502.67 |
| - config_name: read_aloud |
| features: |
| - name: id_recording |
| dtype: string |
| - name: id_sentence |
| dtype: string |
| - name: id_speaker |
| dtype: string |
| - name: text |
| dtype: string |
| - name: location |
| dtype: string |
| - name: location_roomdim |
| dtype: string |
| - name: noise_level |
| dtype: string |
| - name: noise_type |
| dtype: string |
| - name: source_url |
| dtype: string |
| - name: age |
| dtype: int64 |
| - name: gender |
| dtype: string |
| - name: dialect |
| dtype: string |
| - name: country_birth |
| dtype: string |
| - name: validated |
| dtype: string |
| - name: audio |
| dtype: |
| audio: |
| sampling_rate: 16000 |
| - name: asr_prediction |
| dtype: string |
| - name: asr_validation_model |
| dtype: string |
| - name: asr_cer |
| dtype: float64 |
| - name: asr_wer |
| dtype: float64 |
| splits: |
| - name: train |
| num_bytes: 147326850118.0 |
| num_examples: 250108 |
| - name: val |
| num_bytes: 1202472632.0 |
| num_examples: 2046 |
| - name: test |
| num_bytes: 5991057708.0 |
| num_examples: 9123 |
| download_size: 142421368863 |
| dataset_size: 154520380458.0 |
| configs: |
| - config_name: conversation |
| data_files: |
| - split: train |
| path: conversation/train-* |
| - split: val |
| path: conversation/val-* |
| - split: test |
| path: conversation/test-* |
| - config_name: read_aloud |
| data_files: |
| - split: train |
| path: read_aloud/train-* |
| - split: val |
| path: read_aloud/val-* |
| - split: test |
| path: read_aloud/test-* |
| extra_gated_fields: |
| Company/organisation (write N/A if you do not have any affiliation): text |
| Industry: |
| type: select |
| options: |
| - Technology and Services |
| - Healthcare and Life Sciences |
| - Education |
| - Media and Entertainment |
| - Finance and Business |
| - Government and Non-profit |
| - Energy and Environment |
| - Manufacturing and Logistics |
| - Retail and E-commerce |
| - Other industry |
| - Not applicable (e.g., personal use) |
| Intended use: |
| type: select |
| options: |
| - Research and Academic |
| - Commercial Applications |
| - Education and Training |
| - Non-profit and Social Impact |
| - Personal Use |
| - Testing and Benchmarking |
| - label: Other |
| value: Other |
| --- |
| |
|  |
|
|
| # CoRal: Danish Conversational and Read-aloud Dataset |
|
|
| ## Dataset Overview |
|
|
| **CoRal** is a comprehensive Automatic Speech Recognition (ASR) dataset designed to capture the diversity of the Danish language across various dialects, accents, genders, and age groups. The primary goal of the CoRal dataset is to provide a robust resource for training and evaluating ASR models that can understand and transcribe spoken Danish in all its variations. |
|
|
| ### Key Features |
|
|
| - **Dialect and Accent Diversity**: The dataset includes speech samples from all major Danish dialects as well as multiple accents, ensuring broad geographical coverage and the inclusion of regional linguistic features. |
| - **Gender Representation**: Both male and female speakers are well-represented, offering balanced gender diversity. |
| - **Age Range**: The dataset includes speakers from a wide range of age groups, providing a comprehensive resource for age-agnostic ASR model development. |
| - **High-Quality Audio**: All recordings are of high quality, ensuring that the dataset can be used for both training and evaluation of high-performance ASR models. |
|
|
| ### Quick Start |
|
|
| The CoRal dataset is ideal for training ASR models that need to generalise across different dialects and speaker demographics within the Danish language. Below is an example of how to load and use the dataset with Hugging Face's `datasets` library: |
|
|
| ```python |
| from datasets import load_dataset |
| |
| # Load the Coral dataset |
| # Change "read_aloud" to "conversational" to get the conversational dataset |
| coral = load_dataset("CoRal-project/coral-v2", "read_aloud") |
| |
| # Example: Accessing an audio sample and its transcription |
| sample = coral['train'][0] |
| audio = sample['audio'] |
| transcription = sample['text'] |
| |
| print(f"Audio: {audio['array']}") |
| print(f"Text: {transcription}") |
| ``` |
|
|
|
|
| ## Data Fields |
|
|
| - `id_recording`: Unique identifier for the recording. |
| - `id_sentence`: Unique identifier for the text being read aloud. |
| - `id_speaker`: Unique identifier for each speaker. |
| - `text`: Text being read aloud. |
| - `location`: Address of recording place. |
| - `location_roomdim`: Dimension of recording room. |
| - `noise_level`: Noise level in the room, given in dB. |
| - `noise_type`: Noise exposed to the speaker while recording. Note the noise is not |
| present in the audio, but is there to mimic differences in speech in a noisy |
| environment. |
| - `source_url`: URL to the source of the text. |
| - `age`: Age of the speaker. |
| - `gender`: Gender of the speaker. |
| - `dialect`: Self-reported dialect of the speaker. |
| - `country_birth`: Country where the speaker was born. |
| - `validated`: Manual validation state of the recording. |
| - `audio`: The audio file of the recording. |
| - `asr_prediction`: ASR output prediction of the `asr_validation_model`. |
| - `asr_validation_model`: Hugging Face Model ID used for automatic validation of the |
| recordings. |
| - `asr_wer`: Word error rate between `asr_prediction` and `text`. |
| - `asr_cer`: Character error rate between `asr_prediction` and `text`. |
|
|
|
|
| ## Read-aloud Data Statistics |
|
|
| ### Test Split |
|
|
| There are 17.3 hours of audio in the test split, with 35 speakers, reading 9,123 unique sentences aloud. |
|
|
| Gender distribution: |
| - female: 46.8% |
| - male: 53.2% |
|
|
| Dialect and accent distribution: |
| - Bornholmsk: 12.5% |
| - Fynsk: 9.1% |
| - Københavnsk: 9.1% |
| - Nordjysk: 9.0% |
| - Sjællandsk: 7.4% |
| - Sydømål: 6.7% |
| - Sønderjysk: 11.0% |
| - Vestjysk: 8.9% |
| - Østjysk: 9.7% |
| - Non-native accent: 16.7% |
|
|
| Age group distribution: |
| - 0-24: 16.0% |
| - 25-49: 40.1% |
| - 50-: 43.9% |
|
|
| ### Validation Split |
|
|
| There are 3.48 hours of audio in the validation split, with 11 speakers, reading 2046 unique sentences aloud. |
|
|
| Gender distribution: |
| - female: 50.9% |
| - male: 49.1% |
|
|
| Dialect and accent distribution: |
| - Bornholmsk: 8.6% |
| - Fynsk: 11.7% |
| - Københavnsk: 3.9% |
| - Nordjysk: 3.5% |
| - Sjællandsk: 7.1% |
| - Sydømål: 14.1% |
| - Sønderjysk: 5.3% |
| - Vestjysk: 13.7% |
| - Østjysk: 24.7% |
| - Non-native accent: 7.3% |
|
|
| Age group distribution: |
| - 0-24: 34.2% |
| - 25-49: 38.7% |
| - 50-: 27.1% |
|
|
| ### Train Split |
|
|
| There are 425.90 hours of audio in the train split, with 657 speakers, reading 250,108 unique sentences aloud. |
|
|
| Gender distribution: |
| - female: 70.6% |
| - male: 27.5% |
| - non-binary: 2.0% |
|
|
| Dialect distribution: |
| - Bornholmsk: 4.4% |
| - Fynsk: 4.4% |
| - Københavnsk: 13.6% |
| - Nordjysk: 14.7% |
| - Sjællandsk: 15.7% |
| - Sydømål: 0.2% |
| - Sønderjysk: 4.9% |
| - Vestjysk: 11.7% |
| - Østjysk: 26.6% |
| - Non-native accent: 4.0% |
|
|
| Age group distribution: |
| - 0-24: 6.4% |
| - 25-49: 37.2% |
| - 50-: 56.4% |
|
|
|
|
| ## Conversational Data Statistics |
|
|
| More conversational data is expected to released later this year (2025). |
|
|
| ### Test Split |
|
|
| **NOTE: The conversational test split is tentative and is expected to be updated with the next few weeks to be more representative** |
|
|
| There are 1.40 hours of audio in the test split, with 5 speakers. These speakers are all part of the read-aloud test split as well. |
|
|
| Gender distribution: |
| - female: 70.9% |
| - male: 41.9% |
|
|
| Dialect and accent distribution: |
| - Fynsk: 19.2% |
| - Nordjysk: 29.1% |
| - Sønderjysk: 21.2% |
| - Non-native accent: 30.5% |
|
|
| Age group distribution: |
| - 0-24: 0.0% |
| - 25-49: 58.1% |
| - 50-: 41.9% |
|
|
| ### Validation Split |
|
|
| There are 1.16 hours of audio in the validation split, with 4 speakers. These speakers are all part of the read-aloud validation split as well. |
|
|
| Gender distribution: |
| - female: 49.4% |
| - male: 50.6% |
|
|
| Dialect and accent distribution: |
| - Fynsk: 35.7% |
| - Københavnsk: 14.9% |
| - Nordjysk: 35.1% |
| - Østjysk: 14.3% |
|
|
| Age group distribution: |
| - 0-24: 49.9% |
| - 25-49: 0.0% |
| - 50-: 50.1% |
|
|
| ### Train Split |
|
|
| There are 48.85 hours of audio in the train split, with 160 speakers. These speakers are all part of the read-aloud train split as well. |
|
|
| Gender distribution: |
| - female: 78.9% |
| - male: 20.5% |
| - non-binary: 0.6% |
|
|
| Dialect distribution: |
| - Bornholmsk: 1.4% |
| - Fynsk: 3.6% |
| - Københavnsk: 10.4% |
| - Nordjysk: 19.8% |
| - Sjællandsk: 6.8% |
| - Sydømål: 0.0% |
| - Sønderjysk: 4.2% |
| - Vestjysk: 4.9% |
| - Østjysk: 44.9% |
| - Non-native accent: 4.0% |
|
|
| Age group distribution: |
| - 0-24: 5.8% |
| - 25-49: 32.0% |
| - 50-: 62.3% |
|
|
|
|
| ## Example Use Cases |
|
|
| ### ASR Model Training |
|
|
| Train robust ASR models that can handle dialectal variations and diverse speaker demographics in Danish. |
|
|
| ### Dialect Studies |
|
|
| Analyse the linguistic features of different Danish dialects. |
|
|
| ### Forbidden Use Cases |
|
|
| Speech synthesis and biometric identification are not allowed using the CoRal dataset. For more information, see addition 4 in our [license](https://huggingface.co/datasets/alexandrainst/coral/blob/main/LICENSE). |
|
|
|
|
| ## License |
| The dataset is licensed under an OpenRAIL-D license, adapted from OpenRAIL-M, which allows commercial use with a few restrictions (such as speech synthesis and biometric identification). See [license](https://huggingface.co/datasets/alexandrainst/coral/blob/main/LICENSE). |
|
|
|
|
| ## Creators and Funders |
| The CoRal project is funded by the [Danish Innovation Fund](https://innovationsfonden.dk/) and consists of the following partners: |
|
|
| - [Alexandra Institute](https://alexandra.dk/) |
| - [University of Copenhagen](https://www.ku.dk/) |
| - [Agency for Digital Government](https://digst.dk/) |
| - [Alvenir](https://www.alvenir.ai/) |
| - [Corti](https://www.corti.ai/) |
|
|
|
|
| ## Citation |
| We will submit a research paper soon, but until then, if you use the CoRal dataset in your research or development, please cite it as follows: |
|
|
| ```bibtex |
| @dataset{coral2024, |
| author = {Dan Saattrup Nielsen, Sif Bernstorff Lehmann, Simon Leminen Madsen, Anders Jess Pedersen, Anna Katrine van Zee and Torben Blach}, |
| title = {CoRal: A Diverse Danish ASR Dataset Covering Dialects, Accents, Genders, and Age Groups}, |
| year = {2024}, |
| url = {https://hf.co/datasets/alexandrainst/coral}, |
| } |
| ``` |
