Datasets:

deepdml
/

openslr65-tamil

license: cc-by-4.0
language:
  - ta
task_categories:
  - automatic-speech-recognition
tags:
  - audio
  - text
  - speech
  - tamil
  - openslr
pretty_name: OpenSLR-65 Tamil Speech Dataset
size_categories:
  - 1K<n<10K
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*
dataset_info:
  features:
    - name: file_id
      dtype: string
    - name: audio
      dtype: audio
    - name: transcription
      dtype: string
    - name: duration
      dtype: float32
    - name: gender
      dtype: string
  splits:
    - name: train
      num_bytes: 2449492064
      num_examples: 4291
  download_size: 1787306633
  dataset_size: 2449492064

OpenSLR-65 – Tamil Transcribed Speech

Source: https://www.openslr.org/65/

Dataset Description

Homepage: OpenSLR SLR65

This dataset contains transcribed high-quality audio of Tamil sentences recorded by volunteers. It is part of the OpenSLR collection of free speech resources for low-resource languages.

The data was collected via the Appen (formerly Figure Eight / CrowdFlower) crowdsourcing platform and is intended for use in training automatic speech recognition (ASR) and text-to-speech (TTS) systems.

Data Collection

Volunteers were asked to read Tamil sentences displayed on their screen and record themselves. Quality control was performed to ensure accurate transcriptions and clean audio.

Field	Description
`file_id`	Anonymized identifier for the audio file
`transcription`	Tamil text transcription of the utterance
`audio`	WAV audio file (mono)
`duration`	Duration of the audio in seconds
`gender`	Speaker gender (`male` / `female`)

Corpus statistics

gender	samples	duration (h)
female	2 335	4.01
male	1 956	3.07
total	4 291	7.08

Usage

Load the Dataset

from datasets import load_dataset

# Load full dataset
dataset = load_dataset("deepdml/openslr65-tamil")
train_data = dataset["train"]
# train_data = load_dataset("deepdml/openslr65-tamil", split="train")

Inspect a Sample

sample = train_data[0]

print(sample)
# {
#   'file_id':    'tag_09162_01279273055',
#   'audio':      {'array': array([...], dtype=float32)},
#   'transcription': 'அவர்களின் படங்களின் டீஸருக்கு கிடைக்கும் வரவேற்பு அபிரிதமாக உள்ளது',
#   'duration':   5.12,
#   'gender': male,
# }

# Play audio (in a notebook)
import IPython.display as ipd
ipd.Audio(sample["audio"]["array"], rate=sample["audio"]["sampling_rate"])

Filter by Duration

# Keep only utterances between 2 and 15 seconds
filtered = train_data.filter(lambda x: 2.0 <= x["duration"] <= 15.0)

License

Creative Commons Attribution 4.0 International (CC BY 4.0)

Citation

If you use this dataset, please cite:

@inproceedings{he-etal-2020-open,
    title = {{Open-source Multi-speaker Speech Corpora for Building Gujarati, Kannada, Malayalam, Marathi, Tamil and Telugu Speech Synthesis Systems}},
    author = {He, Fei and Chu, Shan-Hui Cathy and Kjartansson, Oddur and Rivera, Clara and Katanova, Anna and Gutkin, Alexander and Demirsahin, Isin and Johny, Cibu and Jansche, Martin and Sarin, Supheakmungkol and Pipatsrisawat, Knot},
    booktitle = {Proceedings of The 12th Language Resources and Evaluation Conference (LREC)},
    month = may,
    year = {2020},
    address = {Marseille, France},
    publisher = {European Language Resources Association (ELRA)},
    pages = {6494--6503},
    url = {https://www.aclweb.org/anthology/2020.lrec-1.800},
    ISBN = "{979-10-95546-34-4},
  }

Additional Information

Homepage: https://www.openslr.org/65/
Repository: https://openslr.org/resources/65/
Paper: N/A
Point of Contact: OpenSLR maintainers