2019 OpenSAT Public Safety Communications Simulation
|Item Name:||2019 OpenSAT Public Safety Communications Simulation|
|Author(s):||Dana Delgado, Karen Jones, Kevin Walker, Stephanie Strassel, Christopher Caruso, David Graff|
|LDC Catalog No.:||LDC2023S06|
|Release Date:||August 15, 2023|
|DCMI Type(s):||Sound, Text|
|Data Source(s):||microphone speech|
|Project(s):||NIST OpenSAT, NIST Public Safety|
|Application(s):||keyword spotting, speech activity detection, speech recognition|
LDC User Agreement for Non-Members
|Online Documentation:||LDC2023S06 Documents|
|Licensing Instructions:||Subscription & Standard Members, and Non-Members|
|Citation:||Delgado, Dana, et al. 2019 OpenSAT Public Safety Communications Simulation LDC2023S06. Web Download. Philadelphia: Linguistic Data Consortium, 2023.|
2019 OpenSAT Public Safety Communications Simulation was developed by the Linguistic Data Consortium (LDC) and contains approximately 141 hours of speech recordings and transcripts used in the used in the National Institute of Standards and Technology (NIST) Open Speech Analytic Technologies (OpenSAT) 2019 evaluation's automatic speech recognition, speech activity detection, and keyword search tasks. The data is a portion of the Speech Analysis For Emergency Response Technology (SAFE-T) corpus, which was created by LDC under the NIST Public Safety project in support of NIST's OpenSAT evaluation campaign.
The NIST OpenSAT evaluation series was designed to bring together researchers developing different types of technologies to address speech analytic challenges present in some of the most difficult acoustic conditions with the end goal of improving the state-of-the-art through objective, large-scale common evaluations. The SAFE-T corpus contains speakers engaged in a collaborative problem-solving activity representative of public safety communications in terms of speech content, noise types and noise levels.
US English speakers played the board game Flash Point Fire Rescue. Background noise was played through a participant's headset during the recording session. Recording sessions consisted of two 30-minute games.
This corpus contains training, development and evaluation data. Development and evaluation audio files consist of four 3-minute snippets selected from the six sections of five minutes each drawn from the 30-minute recording.
All recordings are single channel. The background noise was mixed into the single channel recording at a reduced level. Audio data is presented as 48KHz 16-bit mono flac files. Transcripts are in tab-separated, .tsv format with UTF-8 encoding.
Please view these samples: