Public data

From cslt Wiki
Revision as of 02:48, 28 July 2021 by Lilt (Talk | contribs)

(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to: navigation, search

CCC data resource

CSLT holds a close collaboration with Chinese Corpus Consortium (CCC) to collect and publish databases in China. The aim of the CCC is to provide corpora for Chinese ASR, TTS, NLP, perception analysis, phonetics analysis, linguistic analysis, and other related tasks. The corpora can be speech- or text-based; read or spontaneous; wideband or narrowband; standard or dialectal Chinese; clean or with noise; or of any other kinds which are deemed helpful for the foresaid purposes.

visit CCC


CNCeleb1

This data is a large-scale speaker recognition dataset collected 'in the wild'. The dataset contains more than 130,000 utterances from 1,000 Chinese celebrities, and covers 11 different genres in real world. All the audio files are coded as single channel and sampled at 16kHz with 16-bit precision.

download from openSLR


Trivial events database

A free database involving 7 types of human trivial events: cough, laugh, "wei", "hmm", "tsk-tsk", "ahem", sniff. The data is collected using a recording Android App.

download from Cloud


Disguise database

A free database involving human's normal speech and disguised speech. The data is collected using a recording Android App.

download from Cloud


Uyghur text database

CSLT collaborated with the XinJiang University on a wide range of research including speech recognition, information retrieval and text processing. We published a multitude of resources to boost the research on Uyghur. The text data published here is used for Uyghur text classification tasks, which involves 500 health and non-health documents respectively. It was collected by Mahpirat from XJU when she visited CSLT from 2012-2013.

Download is under construction....


Sheik Cantonese lexicon

A free Cantonese lexicon collected from Adam Sheik's Cantonese Dict project.

check details


THUYG-20 database

A free speech database for constructing a full-fledged Uyghur ASR system.

Download is under construction....


THUYG-20 SRE database

A free speech database for constructing a full-fledged Uyghur speaker recognition system.

Download is under construction....


SUD-12 database

A speech database used for short utterance speaker recognition

check details


THUCH30 database

A speech database used for Chinese LVCSR. Recorded by Dong Wang many many years ago.

check details


Kazak ASR database

A speech database used for Kazak LVCSR.

The entire package involves the full set of speech and language resources required to establish a Kazak speech recognition system.

QQ weiyun share link

You can send e-mail to shiying@cslt.org to ask for share password.


Tibetan ASR database

A speech database used for Tibetan LVCSR.

The entire package involves the full set of speech and language resources required to establish a Tibetan speech recognition system.

QQ weiyun share link

You can send e-mail to shiying@cslt.org to ask for share password.


CSLT-Chronos database

A time-varying dataset for speaker recognition.

The entire package involves the 8 recording sessions selected to evaluate the time-variance effect. For commercial usage, only the MFCCs/Fbanks are published for research usage.

Readme Download

For any query, please contact with lilt@cslt.org


CSLT-ESDB database

A speech dataset for speech emotion recognition.

The entire package involves 4 emotions recorded by 30 actors.

Readme Download

For any query, please contact with shr19@mails.tsinghua.edu.cn