Merge branch 'main' of https://github.com/alibaba-damo-academy/FunASR into main

egs_modelscope/speaker_diarization/speech_diarization_sond-zh-cn-alimeeting-16k-n16k4-pytorch/infer.py

@@ -17,7 +17,7 @@ inference_diar_pipline = pipeline(
     diar_model_config="sond.yaml",
     model='damo/speech_diarization_sond-zh-cn-alimeeting-16k-n16k4-pytorch',
     sv_model="damo/speech_xvector_sv-zh-cn-cnceleb-16k-spk3465-pytorch",
-    sv_model_revision="master",
+    sv_model_revision="v1.2.2",
 )
 
 # use audio_list as the input, where the first one is the record to be detected

funasr/runtime/websocket/readme.md

@@ -51,7 +51,7 @@ make
 
 ```shell
 cd bin
-   ./funasr-wss-server  [--model_thread_num <int>] [--decoder_thread_num <int>]
+./funasr-wss-server  [--model_thread_num <int>] [--decoder_thread_num <int>]
                     [--io_thread_num <int>] [--port <int>] [--listen_ip
                     <string>] [--punc-quant <string>] [--punc-dir <string>]
                     [--vad-quant <string>] [--vad-dir <string>] [--quantize
@@ -85,7 +85,7 @@ Where:
      default: ../../../ssl_key/server.key, path of keyfile for WSS connection
   
 example:
-   funasr-wss-server --model-dir /FunASR/funasr/runtime/onnxruntime/export/damo/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch
+./funasr-wss-server --model-dir /FunASR/funasr/runtime/onnxruntime/export/damo/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch
 ```
 
 ## Run websocket client test

tests/test_sv_inference_pipeline.py

@@ -19,30 +19,20 @@ class TestXVectorInferencePipelines(unittest.TestCase):
             task=Tasks.speaker_verification,
             model='damo/speech_xvector_sv-zh-cn-cnceleb-16k-spk3465-pytorch'
         )
-        # 提取不同句子的说话人嵌入码
-        rec_result = inference_sv_pipline(
-            audio_in='https://isv-data.oss-cn-hangzhou.aliyuncs.com/ics/MaaS/ASR/test_audio/sv_example_enroll.wav')
-        enroll = rec_result["spk_embedding"]
-
-        rec_result = inference_sv_pipline(
-            audio_in='https://isv-data.oss-cn-hangzhou.aliyuncs.com/ics/MaaS/ASR/test_audio/sv_example_same.wav')
-        same = rec_result["spk_embedding"]
-
-        rec_result = inference_sv_pipline(
-            audio_in='https://isv-data.oss-cn-hangzhou.aliyuncs.com/ics/MaaS/ASR/test_audio/sv_example_different.wav')
-        different = rec_result["spk_embedding"]
-
-        # 对相同的说话人计算余弦相似度
-        sv_threshold = 0.9465
-        same_cos = np.sum(enroll * same) / (np.linalg.norm(enroll) * np.linalg.norm(same))
-        same_cos = max(same_cos - sv_threshold, 0.0) / (1.0 - sv_threshold) * 100.0
-        logger.info("Similarity: {}".format(same_cos))
-
-        # 对不同的说话人计算余弦相似度
-        diff_cos = np.sum(enroll * different) / (np.linalg.norm(enroll) * np.linalg.norm(different))
-        diff_cos = max(diff_cos - sv_threshold, 0.0) / (1.0 - sv_threshold) * 100.0
-        logger.info("Similarity: {}".format(diff_cos))
 
+        # the same speaker
+        rec_result = inference_sv_pipline(audio_in=(
+            'https://isv-data.oss-cn-hangzhou.aliyuncs.com/ics/MaaS/ASR/test_audio/sv_example_enroll.wav',
+            'https://isv-data.oss-cn-hangzhou.aliyuncs.com/ics/MaaS/ASR/test_audio/sv_example_same.wav'))
+        assert abs(rec_result["scores"][0]-0.85) < 0.1 and abs(rec_result["scores"][1]-0.14) < 0.1
+        logger.info(f"Similarity {rec_result['scores']}")
+    
+        # different speaker
+        rec_result = inference_sv_pipline(audio_in=(
+            'https://isv-data.oss-cn-hangzhou.aliyuncs.com/ics/MaaS/ASR/test_audio/sv_example_enroll.wav',
+            'https://isv-data.oss-cn-hangzhou.aliyuncs.com/ics/MaaS/ASR/test_audio/sv_example_different.wav'))
+        assert abs(rec_result["scores"][0]-0.0) < 0.1 and abs(rec_result["scores"][1]-1.0) < 0.1
+        logger.info(f"Similarity {rec_result['scores']}")
 
 if __name__ == '__main__':
     unittest.main()