Contact-Centre Conversations

[ OVERVIEW ]

Role-play and consented live contact-centre dialogues designed for training customer-service voice AI, intent classification, and call-handling agents. Every recording is captured with dual-channel separation (agent one channel, customer the other) for clean speaker extraction. Conversations cover billing, cancellations, reservations, complaints, and account changes across retail, finance, travel, and utilities verticals. Every utterance carries an intent label and escalation flag where relevant.

[ KEY HIGHLIGHTS ]

• Dual-channel recording with agent-customer separation
• Intent labels attached per utterance across a taxonomy of 40+ categories
• Escalation and sentiment flags on complaint and resolution turns
• Domain coverage: retail, finance, travel, utilities, and hospitality
• Conversation types: billing, cancellation, reservation, complaint, account-change
• Transcripts include disfluencies and turn-taking markers
• Licensed per-vertical or as the full cross-domain corpus

[ TECHNICAL SPECIFICATIONS ]

Files

Dual-channel WAV, 44.1-48 kHz, 16-bit, with L/R agent-customer separation

Transcripts

JSON with speaker labels, intent tags, word-level timestamps, escalation flags

Annotations

Intent taxonomy · sentiment tagging · escalation severity · domain tag

Licensing

Commercial training rights · per-vertical or full-corpus · custom scoping on request