Working wake word (porcupine) and audio (speech) request to server :)

jarvis/main.py

@@ -0,0 +1,46 @@
+import struct
+
+import pvporcupine
+import pyaudio
+import speech_recognition as sr
+
+from jarvis.utils import server_utils
+
+wake_word_handler = pvporcupine.create(keywords=['jarvis'])
+
+
+def wake_word_listening():
+    pa = pyaudio.PyAudio()
+
+    audio_stream = pa.open(
+        rate=wake_word_handler.sample_rate,
+        channels=1,
+        format=pyaudio.paInt16,
+        input=True,
+        frames_per_buffer=wake_word_handler.frame_length
+    )
+
+    while True:
+        pcm = audio_stream.read(wake_word_handler.frame_length)
+        pcm = struct.unpack_from("h" * wake_word_handler.frame_length, pcm)
+
+        keyword_index = wake_word_handler.process(pcm)
+
+        if keyword_index >= 0:
+            print("Recognized")
+            record()
+
+
+def record():
+    # obtain audio from the microphone
+    r = sr.Recognizer()
+    with sr.Microphone() as source:
+        print("Say something!")
+        r.adjust_for_ambient_noise(source=source, duration=0.5)
+        audio = r.listen(source, timeout=2, phrase_time_limit=5)
+
+    server_utils.send_record_to_server(audio.frame_data)
+
+
+if __name__ == '__main__':
+    wake_word_listening()

jarvis/utils/server_utils.py

@@ -0,0 +1,17 @@
+import requests
+from requests.structures import CaseInsensitiveDict
+
+
+def send_record_to_server(frame_data):
+    # TODO: use config or even ping to find the server on local network ?
+    url_service = "http://192.168.1.12:5000" + "/process_audio_request"
+
+    headers = CaseInsensitiveDict()
+    headers["Content-Type"] = "text/xml; charset=utf8"
+    # headers["Authorization"] = config_utils.get_in_config("API_KEY")
+
+    response = requests.post(url_service,
+                             headers=headers,
+                             data=frame_data)
+
+    print(response.content)

requirements.txt

@@ -0,0 +1,4 @@
+pvporcupine~=1.9.5
+PyAudio~=0.2.11
+SpeechRecognition~=3.8.1
+requests~=2.26.0