InfoBay-AI/call_center_audio_dual_channel_sw

A high-quality call center dual-channel audio dataset (Swahili) containing customer-agent conversations recorded in separate channels. Designed for AI/ML training, speech recognition, speaker diarization, sentiment analysis, call analytics, and conversational intelligence research applications. # call_center_audio_dual_channel_sw **Dataset Description:** This dataset is a **large-scale collection of raw Swahili call center audio (dual-channel)**, designed to support the development and training of advanced speech and AI systems. It consists of real-world customer and agent speech recordings collected from call center environments. The dataset is organized in a dual-channel format, where corresponding customer and agent audio are available as separate recordings, enabling clear role-based modeling and analysis. The dataset captures authentic speech characteristics such as tone variation, pauses, silence patterns, and natural speaking behaviour commonly observed in customer service environments. This makes it highly valuable for building accurate, scalable, and production-ready AI systems for enterprise and customer support applications. Additionally, this dataset can be used in data pipelines for **Supervised Fine-Tuning (SFT)** and **Reinforcement Learning with Human Feedback (RLHF)** workflows. **Key Use Cases** -Training Automatic Speech Recognition (ASR) systems for call center environments -Call analytics and conversation intelligence -Sentiment analysis and emotion detection -Quality monitoring and compliance analysis -Virtual assistants for customer support -Speech-to-Text (STT) for structured call transcription **Dataset Specification** -Language: Swahili -Type: Raw, unprocessed call center audio -Speech Style: Customer and agent speech from call center environments -Audio Conditions: Real-world call center environments -Domains: customer support, service calls, query resolution, etc. -Channel Configuration: Dual-channel (separate customer and agent recordings) -Format: .wav, .mp3, .ogg, etc. -Sampling Rate: 8000 Hz -Duration: Swahili(KE) - 13,426 & Swahili(UG) - 1,753 **Value of Dual Channel Dataset** -Clear separation of customer and agent speech for accurate modeling -Supports role-based speech analysis and modeling -Improves performance …