From 72833a9b7b29d97aadcf4618c30c87bc7a12ff5d Mon Sep 17 00:00:00 2001
From: jiegeaiai <jiegeaiai@163.com>
Date: Thu, 31 Oct 2024 08:15:49 +0800
Subject: [PATCH] add temp commit

---
 human/audio_inference_handler.py | 8 +++++---
 human/audio_mal_handler.py       | 5 ++++-
 2 files changed, 9 insertions(+), 4 deletions(-)

diff --git a/human/audio_inference_handler.py b/human/audio_inference_handler.py
index 187ab73..e43fbcb 100644
--- a/human/audio_inference_handler.py
+++ b/human/audio_inference_handler.py
@@ -34,7 +34,7 @@ class AudioInferenceHandler(AudioHandler):
             self._mal_queue.put(stream)
         elif type_ == 0:
             self._audio_queue.put(stream)
-        print('AudioInferenceHandler on_handle', type_)
+        print('AudioInferenceHandler on_handle', type_, self._audio_queue.size())
 
     def on_message(self, message):
         super().on_message(message)
@@ -70,6 +70,7 @@ class AudioInferenceHandler(AudioHandler):
                 audio_frames = []
                 for _ in range(batch_size * 2):
                     frame, type_ = self._audio_queue.get()
+                    print('AudioInferenceHandler type_', type_)
                     audio_frames.append((frame, type_))
                     if type_ == 0:
                         is_all_silence = False
@@ -129,5 +130,6 @@ class AudioInferenceHandler(AudioHandler):
         self._run_thread.join()
 
     def pause_talk(self):
-        self._audio_queue.clear()
-        self._mal_queue.queue.clear()
+        pass
+        # self._audio_queue.clear()
+        # self._mal_queue.queue.clear()
diff --git a/human/audio_mal_handler.py b/human/audio_mal_handler.py
index 4a8e281..1f295a3 100644
--- a/human/audio_mal_handler.py
+++ b/human/audio_mal_handler.py
@@ -44,15 +44,18 @@ class AudioMalHandler(AudioHandler):
         logging.info('chunk2mal exit')
 
     def _run_step(self):
+        count = 0
         for _ in range(self._context.batch_size * 2):
             frame, _type = self.get_audio_frame()
             self.frames.append(frame)
             self.on_next_handle((frame, _type), 0)
+            print("AudioMalHandler _type", _type)
+            count = count + 1
         # context not enough, do not run network.
         if len(self.frames) <= self._context.stride_left_size + self._context.stride_right_size:
             return
 
-        print('AudioMalHandler _run_step')
+        print('AudioMalHandler _run_step', count)
         inputs = np.concatenate(self.frames)  # [N * chunk]
         mel = melspectrogram(inputs)
         # print(mel.shape[0],mel.shape,len(mel[0]),len(self.frames))