Datasets:

speechcolab
/

gigaspeech

Name: gigaspeech
Creator: SpeechColab
License: https://choosealicense.com/licenses/apache-2.0/

Tasks:

Automatic Speech Recognition

Text-to-Speech

Text-to-Audio

Languages: English

Multilinguality: monolingual

ArXiv:

Tags:

License:

Dataset card Viewer Files Files and versions Community

You need to agree to share your contact information to access this dataset

This repository is publicly accessible, but you have to accept the conditions to access its files and content.

SpeechColab does not own the copyright of the audio files. For researchers and educators who wish to use the audio files for non-commercial research and/or educational purposes, we can provide access through the Hub under certain conditions and terms.
Terms of Access:
The "Researcher" has requested permission to use the GigaSpeech database (the "Database") at Tsinghua University. In exchange for such permission, Researcher hereby agrees to the following terms and conditions:

Researcher shall use the Database only for non-commercial research and educational purposes.
The SpeechColab team and Tsinghua University make no representations or warranties regarding the Database, including but not limited to warranties of non-infringement or fitness for a particular purpose.
Researcher accepts full responsibility for his or her use of the Database and shall defend and indemnify the SpeechColab team and Tsinghua University, including their employees, Trustees, officers and agents, against any and all claims arising from Researcher's use of the Database, including but not limited to Researcher's use of any copies of copyrighted audio files that he or she may create from the Database.
Researcher may provide research associates and colleagues with access to the Database provided that they first agree to be bound by these terms and conditions.
The SpeechColab team and Tsinghua University reserve the right to terminate Researcher's access to the Database at any time.
If Researcher is employed by a for-profit, commercial entity, Researcher's employer shall also be bound by these terms and conditions, and Researcher hereby represents that he or she is fully authorized to enter into this agreement on behalf of such employer.

!!! Please also fill out the Google Form https://forms.gle/UuGQAPyscGRrUMLq6 to request access to the Gigaspeech dataset.

Dataset Card for Gigaspeech

Dataset Description

GigaSpeech is an evolving, multi-domain English speech recognition corpus with 10,000 hours of high quality labeled audio suitable for supervised training. The transcribed audio data is collected from audiobooks, podcasts and YouTube, covering both read and spontaneous speaking styles, and a variety of topics, such as arts, science, sports, etc.

Example Usage

The training split has several configurations of various size: XS, S, M, L, XL. See the Section on "Data Splits" for more information. To download the XS configuration:

from datasets import load_dataset

gs = load_dataset("speechcolab/gigaspeech", "xs", use_auth_token=True)

# see structure
print(gs)

# load audio sample on the fly
audio_input = gs["train"][0]["audio"]  # first decoded audio sample
transcription = gs["train"][0]["text"]  # first transcription

It is possible to download only the development or test data:

gs_dev = load_dataset("speechcolab/gigaspeech", "dev", use_auth_token=True)
gs_test = load_dataset("speechcolab/gigaspeech", "test", use_auth_token=True)

Supported Tasks and Leaderboards

automatic-speech-recognition: The dataset can be used to train a model for Automatic Speech Recognition (ASR). The model is presented with an audio file and asked to transcribe the audio file to written text. The most common evaluation metric is the word error rate (WER). The task has an active leaderboard which can be found at https://github.com/SpeechColab/GigaSpeech#leaderboard and ranks models based on their WER.
text-to-speech, text-to-audio: The dataset can also be used to train a model for Text-To-Speech (TTS).

Languages

Gigaspeech contains audio and transcription data in English.

Dataset Structure

Data Instances

{
    'segment_id': 'YOU0000000315_S0000660', 
    'speaker': 'N/A', 
    'text': "AS THEY'RE LEAVING <COMMA> CAN KASH PULL ZAHRA ASIDE REALLY QUICKLY <QUESTIONMARK>", 
    'audio': 
        {
            # in streaming mode 'path' will be 'xs_chunks_0000/YOU0000000315_S0000660.wav'
            'path': '/home/user/.cache/huggingface/datasets/downloads/extracted/9d48cf31/xs_chunks_0000/YOU0000000315_S0000660.wav', 
            'array': array([0.0005188 , 0.00085449, 0.00012207, ..., 0.00125122, 0.00076294, 0.00036621], dtype=float32), 
            'sampling_rate': 16000
        }, 
    'begin_time': 2941.889892578125, 
    'end_time': 2945.070068359375, 
    'audio_id': 'YOU0000000315', 
    'title': 'Return to Vasselheim | Critical Role: VOX MACHINA | Episode 43', 
    'url': 'https://www.youtube.com/watch?v=zr2n1fLVasU', 
    'source': 2, 
    'category': 24,
    'original_full_path': 'audio/youtube/P0004/YOU0000000315.opus'
}

Data Fields

segment_id (string) - string id of the segment.
speaker (string) - string id of the speaker (can be "N/A").
text (string) - transcription of the segment.
begin_time (float) - start time of the segment in an original full audio.
end_time (float32) - end time of the segment in an original full audio.
audio (Audio feature) - a dictionary containing the path to the audio, the decoded audio array, and the sampling rate. In non-streaming mode (default), the path point to the locally extracted audio. In streaming mode, the path is the relative path of an audio. segment inside its archive (as files are not downloaded and extracted locally).
audio_id (string) - string idea of the original full audio.
title (string) - title of the original full audio.
url (string) - url of the original full audio.
source (ClassLabel) - id of the audio source. Sources are audiobook (0), podcast (1), and YouYube (2).
category (ClassLabel) - id of the audio category, categories are listed below.
original_full_path (string) - the relative path to the original full audio sample in the original data directory.

Categories are assigned from the following labels: "People and Blogs", "Business", "Nonprofits and Activism", "Crime", "History", "Pets and Animals", "News and Politics", "Travel and Events", "Kids and Family", "Leisure", "N/A", "Comedy", "News and Politics", "Sports", "Arts", "Science and Technology", "Autos and Vehicles", "Science and Technology", "People and Blogs", "Music", "Society and Culture", "Education", "Howto and Style", "Film and Animation", "Gaming", "Entertainment", "Travel and Events", "Health and Fitness", "audiobook".

Data Splits

The dataset has three splits: train, evaluation (dev) and test. The train split has five configurations of various sizes: XS, S, M, L, XL. Larger subsets are supersets of smaller subsets, e.g., the L subset contains all the data from the M subset.

Transcribed Training Subsets Size

Subset	Hours	Remarks
XS	10	System building and debugging
S	250	Quick research experiments
M	1,000	Large-scale research experiments
L	2,500	Medium-scale industrial experiments
XL	10,000	Large-scale industrial experiments

Transcribed Evaluation Subsets

Subset	Hours	Remarks
Dev	12	Randomly selected from the crawled Podcast and YouTube Data
Test	40	Part of the subset was randomly selected from the crawled Podcast and YouTube data; part of it was manually collected through other channels to have better coverage.

Dataset Creation

Curation Rationale

[More Information Needed]

Source Data

Audio Source	Transcribed Hours	Acoustic Condition
Audiobook	2,655	Reading Various ages and accents
Podcast	3,498	Clean or background music Indoor Near-field Spontaneous Various ages and accents
YouTube	3,845	Clean and noisy Indoor and outdoor Near- and far-field Reading and spontaneous Various ages and accents
Total	10,000

Initial Data Collection and Normalization

[More Information Needed]

Who are the source language producers?

[More Information Needed]

Annotations

Annotation process

[More Information Needed]

Who are the annotators?

Development and test subsets are annotated by professional human annotators.

Personal and Sensitive Information

[More Information Needed]

Considerations for Using the Data

Social Impact of Dataset

[More Information Needed]

Discussion of Biases

[More Information Needed]

Other Known Limitations

[More Information Needed]

Additional Information

Dataset Curators

[More Information Needed]

Licensing Information

In general, when training a machine learning model on a given dataset, the license of the model is independent to that of the dataset. That is to say, speech recognition models trained on the GigaSpeech dataset may be eligible for commercial license, provided they abide to the 'Fair Use' terms of the underlying data and do not violate any explicit copyright restrictions. This is likely to be true in most use-cases. However, it is your responsiblity to verify the appropriate model license for your specific use-case by confirming that the dataset usage abides by the Fair Use terms. SpeechColab is not responsible for the license of any machine learning model trained on the GigaSpeech dataset.

Citation Information

Please cite this paper if you find this work useful:

@inproceedings{GigaSpeech2021,
  title={GigaSpeech: An Evolving, Multi-domain ASR Corpus with 10,000 Hours of Transcribed Audio},
  booktitle={Proc. Interspeech 2021},
  year=2021,
  author={Guoguo Chen, Shuzhou Chai, Guanbo Wang, Jiayu Du, Wei-Qiang Zhang, Chao Weng, Dan Su, Daniel Povey, Jan Trmal, Junbo Zhang, Mingjie Jin, Sanjeev Khudanpur, Shinji Watanabe, Shuaijiang Zhao, Wei Zou, Xiangang Li, Xuchen Yao, Yongqing Wang, Yujun Wang, Zhao You, Zhiyong Yan}
}

Contributions

Thanks to @polinaeterna and @sanchit-gandhi for adding this dataset.

Terms of Access

The "Researcher" has requested permission to use the GigaSpeech database (the "Database") at Tsinghua University. In exchange for such permission, Researcher hereby agrees to the following terms and conditions:

Researcher shall use the Database only for non-commercial research and educational purposes.
The SpeechColab team and Tsinghua University make no representations or warranties regarding the Database, including but not limited to warranties of non-infringement or fitness for a particular purpose.
Researcher accepts full responsibility for his or her use of the Database and shall defend and indemnify the SpeechColab team and Tsinghua University, including their employees, Trustees, officers and agents, against any and all claims arising from Researcher's use of the Database, including but not limited to Researcher's use of any copies of copyrighted audio files that he or she may create from the Database.
Researcher may provide research associates and colleagues with access to the Database provided that they first agree to be bound by these terms and conditions.
The SpeechColab team and Tsinghua University reserve the right to terminate Researcher's access to the Database at any time.
If Researcher is employed by a for-profit, commercial entity, Researcher's employer shall also be bound by these terms and conditions, and Researcher hereby represents that he or she is fully authorized to enter into this agreement on behalf of such employer.