添加数据集生成工具

honk-master/dataset_generator.py

@@ -0,0 +1,52 @@
+#!/usr/bin/env python3
+import pyaudio
+import argparse
+import hashlib
+import wave
+import os
+def main(flags):
+    os.chdir(flags.p)
+    if not os.path.exists(flags.k):
+        os.mkdir(flags.k)
+    os.chdir(flags.k)
+    RATE=16000
+    CHUNK=1000
+    FORMAT=pyaudio.paInt16
+    RECORD_SECONDS=2
+    _audio=pyaudio.PyAudio()
+    for i in range(0,flags.n):
+        stream = _audio.open(format=pyaudio.paInt16,
+                             channels = 1,
+                             rate=RATE,
+                             input=True,
+                             frames_per_buffer=CHUNK)
+        print("开始录制第%d条语音，请说话"% i)
+        frames = []
+        for j in range(0,int(RATE/CHUNK*RECORD_SECONDS)):
+            data = stream.read(CHUNK)
+            frames.append(data)
+        print("录音结束")
+        stream.stop_stream()
+        stream.close()
+        hashcode=hashlib.sha224(b''.join(frames)).hexdigest()[0:8]
+        wf=wave.open(hashcode+"__nohash__.wav",'wb')
+        wf.setnchannels(1)
+        wf.setsampwidth(_audio.get_sample_size(FORMAT))
+        wf.setframerate(RATE)
+        wf.writeframes(b''.join(frames))
+        wf.close()
+        print("保存完毕,请按回车进行下一轮录制")
+        str=input();
+
+
+if __name__ == "__main__":
+    parser = argparse.ArgumentParser()
+    parser.add_argument("--k",type=str,default="",help="the keyword")
+    parser.add_argument("--p",type=str,default=".",help="the path of dataset")
+    parser.add_argument("--n",type=int,default=10,help="the number of wav files")
+    flags = parser.parse_args()
+    if flags.k=="":
+        print("keyword can not be empty!" )
+        exit();
+    main(flags)    
+