From 79df82ebea8494007f13bcdfedc9372132143baf Mon Sep 17 00:00:00 2001
From: Yun <2289128964@qq.com>
Date: Thu, 4 Jul 2024 09:46:42 +0800
Subject: [PATCH] =?UTF-8?q?feat:=20=E5=AE=8C=E5=96=84=E4=BF=AE=E6=94=B9?=
 =?UTF-8?q?=E6=88=90=E8=87=AA=E5=8A=A8=E7=BB=9D=E5=AF=B9=E8=B7=AF=E5=BE=84?=
 =?UTF-8?q?,=E6=B7=BB=E5=8A=A0=E6=8E=A5=E5=8F=A3=E7=94=9F=E6=88=90?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

---
 baseasr.py | 61 ++++++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 61 insertions(+)
 create mode 100644 baseasr.py

diff --git a/baseasr.py b/baseasr.py
new file mode 100644
index 0000000..df66873
--- /dev/null
+++ b/baseasr.py
@@ -0,0 +1,61 @@
+import time
+import numpy as np
+
+import queue
+from queue import Queue
+import multiprocessing as mp
+
+
+class BaseASR:
+    def __init__(self, opt):
+        self.opt = opt
+
+        self.fps = opt.fps # 20 ms per frame
+        self.sample_rate = 16000
+        self.chunk = self.sample_rate // self.fps # 320 samples per chunk (20ms * 16000 / 1000)
+        self.queue = Queue()
+        self.output_queue = mp.Queue()
+
+        self.batch_size = opt.batch_size
+
+        self.frames = []
+        self.stride_left_size = opt.l
+        self.stride_right_size = opt.r
+        #self.context_size = 10
+        self.feat_queue = mp.Queue(2)
+
+        #self.warm_up()
+
+    def pause_talk(self):
+        self.queue.queue.clear()
+
+    def put_audio_frame(self,audio_chunk): #16khz 20ms pcm
+        self.queue.put(audio_chunk)
+
+    def get_audio_frame(self):        
+        try:
+            frame = self.queue.get(block=True,timeout=0.01)
+            type = 0
+            #print(f'[INFO] get frame {frame.shape}')
+        except queue.Empty:
+            frame = np.zeros(self.chunk, dtype=np.float32)
+            type = 1
+
+        return frame,type 
+
+    def get_audio_out(self):  #get origin audio pcm to nerf
+        return self.output_queue.get()
+    
+    def warm_up(self):
+        for _ in range(self.stride_left_size + self.stride_right_size):
+            audio_frame,type=self.get_audio_frame()
+            self.frames.append(audio_frame)
+            self.output_queue.put((audio_frame,type))
+        for _ in range(self.stride_left_size):
+            self.output_queue.get()
+
+    def run_step(self):
+        pass
+
+    def get_next_feat(self,block,timeout):        
+        return self.feat_queue.get(block,timeout)
\ No newline at end of file