CoSE-E: A Benchmark for Code-switched Speech Evaluation in Enterprise Settings
The paper measures code-switching ASR by whether transcription errors break enterprise voice-agent tasks.
CoSE-E is a synthetic benchmark and evaluation framework for multilingual enterprise speech settings. The authors test frontier ASR systems across five language pairs and analyze the extra errors introduced by code-switching. They frame the release as support for evaluating multilingual voice agents before deployment. ArXiv · AI/CL/LG's note
CoSE-E is a synthetic benchmark and evaluation framework for multilingual enterprise speech settings. The authors test frontier ASR systems across five language pairs and analyze the extra errors introduced by code-switching. They frame the release as support for evaluating multilingual voice agents before deployment. ArXiv · AI/CL/LG's note
score 5