AppTek Call-Center Dialogues: 128.6 Hours of Multi-Accent English Speech

128.6 hours of long-form conversational speech data designed as a benchmark for automatic speech recognition. The dataset features diverse English accents across 16 service domains, with conversations lasting 5–15 minutes. It was created by apptek-com and last updated on the Hugging Face platform in April 2026.

Use Cases

• Benchmarking ASR model performance on realistic call-center interactions based on the long-form conversational structure.
• Training accent-robust speech recognition systems based on the 14 English accent groups.
• Developing domain-specific language models for customer service based on the 16 service domains mentioned.
• Evaluating speaker diarization or channel separation techniques based on the split-channel audio format.

Strengths

• Contains 128.6 hours of speech data, providing substantial material for model training and evaluation.
• Features 14 distinct English accent groups, supporting research into accent robustness.
• Covers 16 different service-oriented domains, increasing the dataset's applicability to real-world scenarios.
• Consists of long-form conversations (5–15 minutes), which is valuable for testing models on extended dialogue.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment for specific modeling tasks.
• The dataset's specific license is unknown, which could restrict usage.

Provenance

Source

apptek-com via Hugging Face

Collection Method

Likely collected from call-center interactions, but the exact gathering method is unspecified.

Freshness

Last updated 2026-04-29 19:18:48; freshness should be verified.

Geography

Global, based on the inclusion of multiple English accents, but specific geographic coverage is not detailed.