Fisher English Training Part 2, Transcripts

Introduction

Fisher English Training Part 2, Transcripts was developed by the Linguistic Data Consortium (LDC) and contains time-aligned transcripts for approximately 975 hours of English conversational telephone speech (CTS).

The corresponding speech files for these transcripts are available in Fisher English Training Part 2, Speech (LDC2005S13).

These two corpora represent the second half of a CTS collection that was created at LDC during 2003. The first half of the collection, released in 2004, comprises Fisher English Training Speech Part 1 Speech (LDC2004S13) and Fisher English Training Speech Part 1 Transcripts (LDC2004T19). Taken as a whole, the two parts contain 11,699 recorded telephone conversations totalling approximately 1,960 hours.

The Fisher telephone conversation collection protocol was created at LDC to address a critical need of developers trying to build robust automatic speech recognition (ASR) systems. Previous collection protocols, such as CALLFRIEND and Switchboard-II and the resulting corpora, have been adapted for ASR research but were in fact developed for language and speaker identification respectively. Although the CALLHOME protocol and corpora were developed to support ASR technology, they feature small numbers of speakers making telephone calls of relatively long duration with narrow vocabulary across the collection. CALLHOME conversations are challengingly natural and intimate. The Fisher protocol uses a large number of participants, and each one converses with another participant, whom they typically do not know, for a short period of time to discuss the assigned topics. This maximizes inter-speaker variation and vocabulary breadth while also increasing formality.

Previous protocols such as CALLHOME, CALLFRIEND, and Switchboard relied on participant activity to drive the collection. Fisher is unique in being platform-driven rather than participant-driven. Participants who wish to initiate a call may do so, however, the collection platform initiates the majority of calls. Participants need only answer their phones at the times they specified when registering for the study.

To encourage a broad range of vocabulary, Fisher participants are asked to speak on an assigned topic randomly selected from a list, which changes every 24 hours and is assigned to all subjects paired on that day. Some topics are inherited or refined from previous Switchboard studies while others were developed specifically for the Fisher protocol.

Data

The individual audio files are presented in NIST SPHERE format, and contain two-channel mu-law sample data. Shorten compression has been applied to all files. Transcription files are stored in .txt format, with speaker turns and timestamps included.

For the entire collection, here is the gender breakdown for the participants: 6,813 females, 5,104 males.

About 12% of the conversations were transcribed by LDC using automatic segmentation and then creating a timestamped text file that was filled in with manual transcription. The rest of the conversations were transcribed by BBN and WordWave by creating a manual transcription of the entire conversation and then sending the audio and text through a forced alignment system to create the timestamps. The original BBN output consisted of five files, the original full file with no timestamps and then four separate outputs from the forced-alignment process. In this release, the BBN output was normalized to match the formatting of the LDC files. The "bbn_orig" directory includes just the transcript data created by BBN, not including the files transcribed by LDC, and the "trans" directory contains all the conversations rendered in LDC formatting.

Data collection and transcription were sponsored by DARPA and the U.S. Department of Defense, as part of the EARS project for research and development in ASR.

Samples

For an example of the data in this corpus, please view this sample (TXT).

Updates

None at this time.