diff --git a/README.md b/README.md index 6da4ef2..99285fe 100644 --- a/README.md +++ b/README.md @@ -159,5 +159,28 @@ git clone https://www.modelscope.cn/iic/SenseVoiceSmall.git Undo +# Performance + +## 语音识别效果 +我们在开源基准数据集(包括 AISHELL-1、AISHELL-2、Wenetspeech、Librispeech和Common Voice)上比较了SenseVoice与Whisper的多语言语音识别性能和推理效率。在中文和粤语识别效果上,SenseVoice-Small模型具有明显的效果优势。 + +
+
+
+
+
如果您是本模型的贡献者,我们邀请您根据模型贡献文档,及时完善模型卡片内容。
diff --git a/fig/asr_results.png b/fig/asr_results.png new file mode 100644 index 0000000..d962606 Binary files /dev/null and b/fig/asr_results.png differ diff --git a/fig/inference.png b/fig/inference.png new file mode 100644 index 0000000..dd9d3a7 Binary files /dev/null and b/fig/inference.png differ