Dev gzf (#1435)

* qwenaudio qwenaudiochat

* qwenaudio qwenaudiochat

Author: LauraGPT

examples/industrial_data_pretraining/qwen_audio/demo.py

@@ -9,5 +9,8 @@
 
 model = AutoModel(model="Qwen-Audio")
 
-res = model.generate(input="https://isv-data.oss-cn-hangzhou.aliyuncs.com/ics/MaaS/ASR/test_audio/asr_example_zh.wav", language=None)
+audio_in = "https://github.com/QwenLM/Qwen-Audio/raw/main/assets/audio/1272-128104-0000.flac"
+prompt = "<|startoftranscription|><|en|><|transcribe|><|en|><|notimestamps|><|wo_itn|>"
+
+res = model.generate(input=audio_in, prompt=prompt)
 print(res)

examples/industrial_data_pretraining/qwen_audio/demo_from_local.py

@@ -11,5 +11,8 @@
                   model_path="/nfs/zhifu.gzf/init_model/qwen/Qwen-Audio",
                   )
 
-res = model.generate(input="https://isv-data.oss-cn-hangzhou.aliyuncs.com/ics/MaaS/ASR/test_audio/asr_example_zh.wav", language=None)
+audio_in = "https://github.com/QwenLM/Qwen-Audio/raw/main/assets/audio/1272-128104-0000.flac"
+prompt = "<|startoftranscription|><|en|><|transcribe|><|en|><|notimestamps|><|wo_itn|>"
+
+res = model.generate(input=audio_in, prompt=prompt)
 print(res)

funasr/models/qwen_audio/model.py

@@ -54,9 +54,8 @@ def inference(self,
 
         meta_data = {}
         # meta_data["batch_data_time"] = -1
-
-        sp_prompt = "<|startoftranscription|><|en|><|transcribe|><|en|><|notimestamps|><|wo_itn|>"
-        query = f"<audio>{data_in[0]}</audio>{sp_prompt}"
+        prompt = kwargs.get("prompt", "<|startoftranscription|><|en|><|transcribe|><|en|><|notimestamps|><|wo_itn|>")
+        query = f"<audio>{data_in[0]}</audio>{prompt}"
         audio_info = self.tokenizer.process_audio(query)
         inputs = self.tokenizer(query, return_tensors='pt', audio_info=audio_info)
         inputs = inputs.to(self.model.device)