What problem does it solve? Evaluating automatic speech recognition (ASR) model inference on Hygon DCU hardware requires manual container setup, dataset mounting, and metric collection. This Skill automates the full evaluation workflow for the SenseVoiceSmall multilingual speech recognition model, from Docker container launch to CER/WER accuracy reporting. ## Core Features & Use Cases - Containerized DCU Evaluation: Launches a preconfigured Docker environment with Hygon DTK 25.04.2, mounting model weights, LMDB datasets, and configs with correct device and permission flags. - Multi-Dataset ASR Benchmarking: Runs inference across 28 speech recognition datasets including Aishell-1, LibriSpeech, FLEURS, and KeSpeech dialect variants. - Automated Metric Collection: Extracts CER (character error rate), WER (word error rate), average inference time, and success rate from acc_report.json into a structured result.json. - Use Case: A hardware validation engineer needs to verify SenseVoiceSmall inference accuracy and throughput on a Hygon DCU cluster; this Skill guides the executor through container startup, dataset selection, inference execution, and metric extraction in one flow. ## Quick Start Ask the agent to test SenseVoiceSmall speech recognition inference performance on Hygon DCU using the Aishell-1 dataset and collect the CER metrics.
