diff --git a/sounddevice_client.py b/sounddevice_client.py
new file mode 100644
index 0000000..66a1adb
--- /dev/null
+++ b/sounddevice_client.py
@@ -0,0 +1,52 @@
+"""
+Simple (hopefully) multiplatform client mimicing the linux command
+
+    arecord -f S16_LE -c1 -r 16000 -t raw -D default | nc localhost 43001
+
+which streams audio data from the microphone to the server.
+Tested on Mac Os.
+"""
+import sys
+import sounddevice as sd
+import socket
+import numpy as np
+from argparse import ArgumentParser
+
+
+parser = ArgumentParser(__doc__)
+parser.add_argument('--host', type=str, default='localhost', help='Host name')
+parser.add_argument('--port', type=int, default=43007, help='Port number')
+parser.add_argument('--chunk', type=int, default=1000, help='Chunk size in ms')
+args = parser.parse_args()
+
+
+
+# Audio configuration needed for whisper_online_server.py
+SAMPLE_RATE = 16000
+CHANNELS = 1
+CHUNK = 1024
+# SIGNED INT16 LITTLE ENDIAN is setup as for sounddevice
+DTYPE = np.int16
+
+sock = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
+sock.connect((args.host, args.port))
+
+print("Recording and streaming audio...")
+
+def callback(indata, frames, time, status):
+    if status:
+        print(status, file=sys.stderr)
+    # Convert the audio data to bytes and send it over the socket
+    sock.sendall(indata.tobytes())
+
+try:
+    # Open the audio stream
+    with sd.InputStream(samplerate=SAMPLE_RATE, channels=CHANNELS, dtype=DTYPE, callback=callback, blocksize=CHUNK):
+        print("Press Ctrl+C to stop the recording")
+        while True:
+            sd.sleep(args.chunk)
+except KeyboardInterrupt:
+    print("Stopping...")
+finally:
+    # Close the socket
+    sock.close()