CALLHOME Japanese Second Edition
| Item Name: | CALLHOME Japanese Second Edition |
| Author(s): | Alexandra Canavan, George Zipperlen, Barbara Wheatley, Masayo Kaneko, Megumi Kobayashi, Neville Ryant, Danni Ma |
| LDC Catalog No.: | LDC2026S02 |
| ISLRN: | 919-730-311-484-7 |
| DOI: | https://doi.org/10.35111/65h8-rh27 |
| Release Date: | January 15, 2026 |
| Member Year(s): | 2026 |
| DCMI Type(s): | Sound, Text |
| Sample Type: | 16-bit FLAC |
| Sample Rate: | 8000 |
| Data Source(s): | telephone conversations |
| Project(s): | Hub5-LVCSR |
| Application(s): | speaker identification, speech recognition |
| Language(s): | Japanese |
| Language ID(s): | jpn |
| License(s): |
LDC User Agreement for Non-Members |
| Online Documentation: | LDC2026S02 Documents |
| Licensing Instructions: | Subscription & Standard Members, and Non-Members |
| Citation: | Canavan, Alexandra, et al. CALLHOME Japanese Second Edition LDC2026S02. Web Download. Philadelphia: Linguistic Data Consortium, 2026. |
| Related Works: | View |
Introduction
CALLHOME Japanese Second Edition was developed by the Linguistic Data Consortium (LDC) and contains approximately 49 hours of speech from 120 unscripted telephone conversations between native Japanese speakers. This publication is a re-release of the original CALLHOME Japanese collection, combining CALLHOME Japanese Speech (LDC96S37) and CALLHOME Japanese Transcripts (LDC96T18), with additional transcription and updated directory structure, file formats, and documentation.
The CALLHOME series consists of telephone conversations and transcripts developed by LDC and Rutgers, The State University of New Jersey, in support of research in speaker identification, language identification and related technologies. Languages in the series include American English, Egyptian Arabic, German, Japanese, Mandarin Chinese, and Spanish.
Data
This release contains the 120 telephone conversations published in CALLHOME Japanese Speech (LDC96S37) which represented training and development data and a subset of evaluation data. Calls originated in North America and were placed to locations overseas. Most participants called family members or close friends. Participants spoke on topics of their choice in a single telephone call lasting up to 30 minutes. Calls were manually audited for language, recording quality, channel characteristics, dialect, and region.
The audio was originally recorded as 8 kHz u-law SPHERE files compressed with SHORTEN. For this second edition, all audio was converted to FLAC format. Recordings are provided as 8 kHz, 16-bit, two-channel FLAC files. The original training/development/test partitioning was removed. All files appear in a unified directory.
Transcripts are provided as UTF-8 encoded TSV files in WebTrans format (LDC's standard transcription tool). Two versions are included: (1) the transcripts published in CALLHOME Japanese Transcripts (LDC96T18); and (2) revised transcripts conforming to updated LDC transcription guidelines. The latter includes normalization of annotation formats, standardization of speaker-produced and background noises, application of foreign-language marking, whitespace cleanup, and corrections and consistency fixes.
Provided metadata includes call-level information (background noise, distortion, crosstalk), speaker metadata (accent, age, sex, comments), and demographic information for call initiators (age, education level).
Samples
Updates
No Updates at this time.