| --- |
| license: mit |
| --- |
| --- |
| # For reference on dataset card metadata, see the spec: https://github.com/huggingface/hub-docs/blob/main/datasetcard.md?plain=1 |
| # Doc / guide: https://huggingface.co/docs/hub/datasets-cards |
| {} |
| --- |
|
|
| # Dataset Card for Dataset proverbs |
|
|
| <!-- Provide a quick summary of the dataset. --> |
|
|
| Audiofiles of reading proverbs in Russian. |
|
|
| ## Dataset Details |
|
|
| ### Dataset Description |
|
|
| The dataset consist of audiofiles proverbs read by children by memory. The text can be slightly different from the real prononsiation of the child, because of prononsiation mistakes. |
|
|
|
|
|
|
| - **Curated by:** Pavel Roudych |
| - **Language(s) (NLP):** Russian |
| - **License:** MIT |
|
|
| ## Uses |
|
|
| <!-- Address questions around how the dataset is intended to be used. --> |
| The dataset can be used to finetune the models to increase the recognition of children speach. |
|
|
|
|
|
|
| ### Source Data |
|
|
| The dataset consist of custom recorded and downloaded from the open sourcec children reading speach. |
|
|
| #### Data Collection and Processing |
|
|
| The records were manually recorded or downloaded, the showed texts were manually allocated and veryfied. The records were resampled and repacked into 16kHz wav. |
