dataset_info:
features:
- name: audio
dtype:
audio:
sampling_rate: 16000
- name: sentence
dtype: string
splits:
- name: train
num_bytes: 9297874494.784
num_examples: 76356
- name: test
num_bytes: 573909907.992
num_examples: 4314
- name: validation
num_bytes: 1092954767.768
num_examples: 7974
download_size: 10936659020
dataset_size: 10964739170.544
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: test
path: data/test-*
- split: validation
path: data/validation-*
Dataset Card for Kannada ASR Dataset (16kHz)
This dataset card provides detailed information about the Kannada ASR Dataset, which is designed for Automatic Speech Recognition tasks.
Dataset Details
Dataset Description
The Kannada ASR Dataset contains audio files and their corresponding transcriptions, specifically curated for training ASR models. The audio files are resampled to 16kHz.
- Curated by: Md Raqeeb Haider
- Language(s) (NLP): Kannada
Dataset Sources
- Repository: haideraqeeb/kannada-asr-16kHz
- Source : SpeechLabs IITM
Uses
Direct Use
The dataset is intended for training and evaluating ASR models in the Kannada language.
Out-of-Scope Use
The dataset is not suitable for tasks unrelated to speech recognition or those requiring languages other than Kannada.
Dataset Structure
The dataset is divided into three splits: train, test, and validation. Each split contains audio files in .wav format and their corresponding transcriptions.
Dataset Creation
Curation Rationale
The dataset was created to support the development of ASR systems for the Kannada language, which is underrepresented in existing datasets.
Source Data
Data Collection and Processing
This dataset was created by getting data from SpeechLabs IITM website after which it was preprocessed so that it can be used directly for ASR pipelines.
Recommendations
Users should consider augmenting the dataset with additional data to mitigate potential biases and improve model robustness.
More Information
For more details, please contact Md Raqeeb Haider at [[email protected]].
Dataset Card Authors
Md Raqeeb Haider
Dataset Card Contact
Md Raqeeb Haider - [[email protected]]