HKUST Mandarin Telephone Speech, Part 1
|Item Name:||HKUST Mandarin Telephone Speech, Part 1|
|Author(s):||Pascale Fung, Shudong Huang, David Graff|
|LDC Catalog No.:||LDC2005S15|
|Release Date:||July 15, 2005|
|Data Source(s):||telephone speech|
|Application(s):||automatic content extraction|
LDC User Agreement for Non-Members
|Online Documentation:||LDC2005S15 Documents|
|Licensing Instructions:||Subscription & Standard Members, and Non-Members|
|Citation:||Fung, Pascale, Shudong Huang, and David Graff. HKUST Mandarin Telephone Speech, Part 1 LDC2005S15. Web Download. Philadelphia: Linguistic Data Consortium, 2005.|
HKUST Mandarin Telephone Speech, Part 1 was developed by Hong Kong University of Science and Technology (HKUST) and contains approximately 149 hours of conversational telephone speech (CTS) in Mandarin. Given that Standard Mandarin is not the native dialect in many regions of China but is the official language of education, speakers may or may not have regional accents speaking Mandarin. It was decided that subjects' birthplaces were divided into Mandarin-dominant and non-Mandarin-dominant regions and all calls were audited and classified into standard and accented types without further distinctions.
In 2004, HKUST was contracted to collect and transcribe 200 hours of Mandarin Chinese CTS from Mandarin speakers in mainland China under the DARPA EARS framework. The first 50 hours of speech and transcripts were released in June 2004 to the EARS community for the RT-04 NIST evaluation. NIST partitioned the remaining 150 hours of collection into training, development, and evaluation sets. This release contains the training and development sets.
The corresponding transcripts for these speech files are available in HKUST Mandarin Telephone Transcript Data, Part 1 (LDC2005T32).
Subject recruitment was done in several cities across mainland China. Most subjects did not previously know each other. To encourage more meaningful conversation, topics similar to those in Fisher English were designed. All calls were initiated by an automated operator calling two participants as scheduled to initiate a call. Subjects were asked about demographic questions (gender, age, native language/dialect, birthplace, education, occupation, phone type, etc.) before they were bridged for normal conversation. Their answers to select demographic questions are part of the call list files in the corpus.
Subjects were allowed to talk up to 10 minutes. With a few exceptions, most calls are of the maximum length. Although subjects were allowed to make up to three calls, all subjects made just one call in this release with one exception, where PIN 10683 and PIN 10686 belong to a single individual. Here's a breakdown of the quantities and gender distribution for the calls by set:
Each call side was recorded on a separate .wav file, sampled at 8-bits (a-law encoded), 8 kHz. They were multiplexed later in sphere format with a-law encoding preserved. In the case where one side was shorter than the other, the shorter side was padded with silence. In the release, the file name of each recorded call is in the format of date_time_Apin_Bpin.sph and the corresponding transcript in LDC2005T32 is in the same format with .txt extension.
For an example of the data in this corpus, please listen to these audio samples: WAV or MP3.
None at this time.