CALLHOME Japanese: 120 Telephone Conversations for Speech Research

Approximately 49 hours of speech from 120 unscripted telephone conversations between native Japanese speakers, developed by the Linguistic Data Consortium (LDC). This second edition re-releases the original CALLHOME Japanese collection with updated transcripts, file formats, and documentation. The data was collected as part of the CALLHOME series to support research in speaker and language identification.

Use Cases

• Train speaker identification models based on unscripted conversational audio.
• Develop language identification systems using Japanese telephone speech.
• Analyze conversational speech patterns and topics from family and friend calls.
• Benchmark automatic speech recognition systems on spontaneous Japanese dialogue.
• Study the normalization of transcription formats and annotation standards.

Strengths

• Contains approximately 49 hours of speech from 120 distinct conversations.
• Provides two versions of transcripts: original and revised with updated LDC guidelines.
• Includes metadata on call-level factors (noise, distortion) and speaker demographics (age, sex, accent).

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• The original training/development/test partitioning has been removed, which may complicate reproducibility for some research.
• Row count and exact file size are unknown, limiting suitability assessment.

Provenance

Source

Linguistic Data Consortium (LDC) and Rutgers, The State University of New Jersey.

Collection Method

Manually audited telephone conversations recorded between participants in North America and overseas locations.

Freshness

Last updated 2026-04-25 08:10:12; freshness should be verified.

Geography

Calls originated in North America and were placed to locations overseas.