Switchboard-300hr

The 'Switchboard-300hr' dataset contains 300 hours of conversational telephone speech and is used to evaluate automatic speech recognition (ASR) systems' performance, specifically their word error rates (WER).

Papers using Switchboard-300hr (6)

Right Label Context in End-to-End Training of Time-Synchronous ASR Models2025Improving the Training Recipe for a Robust Conformer-based Hybrid Model2022Confidence Score Based Speaker Adaptation of Conformer Speech Recognition Systems2023Competitive and Resource Efficient Factored Hybrid HMM Systems are Simpler Than You Think2023One-pass Multiple Conformer and Foundation Speech Systems Compression and Quantization Using An All-in-one Neural Model2024Investigating the Effect of Label Topology and Training Criterion on ASR Performance and Alignment Quality2024