feat(eval): ASR 模型评估框架

横向对比云端 gummy 与本地开源模型(faster-whisper/SenseVoice/Paraformer)，
重点覆盖中英混说，产出准确率(CER/WER/MER)/速度(延迟/RTF)/资源(cpu/mem/模型大小)
对比报告。公共集(ASCEND/AISHELL/LibriSpeech)统一走 HF 适配器 + 自定义 JSONL manifest。
gummy 引擎对照 server/internal/asr/gummy.go 协议移植。

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

This commit is contained in:

wangjia

2026-06-13 11:25:04 +08:00

parent 40760aa884

commit 25acf9db6e

25 changed files with 1594 additions and 0 deletions

									
										eval/asr_eval/__main__.py
									
		+4
		
												View File
												
				@@ -0,0 +1,4 @@

				from asr_eval.cli import main

				if __name__ == "__main__":

				    main()