metadata
dataset_info:
- config_name: dubs
features:
- name: audio
dtype: audio
- name: text
dtype: string
- name: duration
dtype: float64
- name: id
dtype: string
- name: file_id
dtype: string
- name: source
dtype: string
- name: speaker_id
dtype: string
splits:
- name: train
configs:
- config_name: dubs
data_files:
- split: train
path: '*/train*.parquet'
language:
- tt
pretty_name: Tatar Dubs
task_categories:
- text-to-speech
- automatic-speech-recognition
- audio-to-audio
tags:
- audio
- tts
- speech
- speech-to-speech
size_categories:
- 1K<n<10K
license: cc-by-4.0
Studio Dub Recordings
Dataset Summary
Studio Dub Recordings is a speech dataset consisting of high-quality studio recordings of dubbed audio in Tatar language. The dataset features professional voice actors performing various scripted dialogues, intended primarily for speech and text-to-speech (TTS) research in Tatar.
Dataset Structure
Parts:
The dataset contains two main configurations:
audiobooks: Contains the original audiobooks recordings.
Data Fields:
audio: The audio file containing the spoken fairytale segment.text: The corresponding transcription.duration: The duration of the audio segment in seconds.id: A unique identifier for each segment within a file.file_id: A unique identifier for each part of the book.source: The title of the book.speaker_id: The id of each reader.
Data Processing
Text Processing:
- All the text was splited, using razdel library.
Audio Processing:
- All audio was aligned to the text using ctc-forced-aligner library with default model.
Technical Details:
| Dataset Type | speech corpus for TTS |
| Language | tat/tt, Tatar |
| Speech Style | scripted dialogues |
| Content | film dubbings |
| Audio Parameters | 44.1 kHz, 32 bits, mono |
| File Format | WAV (PCM) TXT (UTF-8) |
| Recording Environment | quiet indoor environment |
Total Duration:
Dubs: 4:36:38 (4 hours, 36 minutes, 38 seconds)
Usage Considerations
- The dataset was not extensively preprocessed. Users are encouraged to perform additional preprocessing (e.g., normalization, cleaning, or re-segmentation) as needed for their specific applications.
- All texts are written in Cyrillic script.
Citation
If you use this dataset in your work, please, cite us.
Acknowledgments
This dataset is created with support from the community project Безнең Тәҗрибә. To support their work, please visit: https://boosty.to/beztej
License
This dataset is distributed under the CC BY-SA 4.0 license.