Added arecord audio system and settings

rhasspy/app.py

@@ -16,7 +16,6 @@
 
 from flask import Flask, request, Response, jsonify, send_file, send_from_directory
 from flask_cors import CORS
-import pyaudio
 import requests
 
 import utils
@@ -25,6 +24,7 @@
 from stt import transcribe_wav, maybe_load_decoder
 from intent import best_intent
 from train import train
+from audio_recorder import PyAudioRecorder, ARecordAudioRecorder
 
 # -----------------------------------------------------------------------------
 
@@ -547,52 +547,30 @@ def speech_to_intent(profile, wav_data, no_hass=False):
 
 # -----------------------------------------------------------------------------
 
-record_audio = None
-record_mic = None
-record_buffer = None
+recorder = None
 
-# Use arecord to start recording a WAV file to a temporary file
+# Start recording a WAV file to a temporary buffer
 @app.route('/api/start-recording', methods=['POST'])
 def api_start_recording():
-    global record_audio, record_mic, record_buffer
+    global recorder
     device_index = int(request.args.get('device', -1))
     if device_index < 0:
         device_index = None  # default device
 
-    def stream_callback(data, frame_count, time_info, status):
-        global record_buffer
-        if record_buffer is None:
-            record_buffer = data
-        else:
-            record_buffer += data
-
-        return (data, pyaudio.paContinue)
-
-    record_audio = pyaudio.PyAudio()
-    data_format = record_audio.get_format_from_width(2)  # 16-bit
-    record_mic = record_audio.open(format=data_format,
-                                   channels=1,
-                                   rate=16000,
-                                   input_device_index=device_index,
-                                   input=True,
-                                   stream_callback=stream_callback)
-
-    record_mic.start_stream()
+    profile = request_to_profile(request, profiles_dirs)
+    recorder = get_audio_recorder(profile)
+    recorder.start_recording(device_index)
 
     return 'OK'
 
 # Stop recording WAV file, transcribe, and get intent
 @app.route('/api/stop-recording', methods=['POST'])
 def api_stop_recording():
-    global decoders, record_audio, record_mic, record_buffer
+    global decoders, recorder
     no_hass = request.args.get('nohass', 'false').lower() == 'true'
 
-    if (record_audio is not None) and (record_mic is not None) and (record_buffer is not None):
-        record_mic.stop_stream()
-        record_audio.terminate()
-
-        record_audio = None
-        record_mic = None
+    if recorder is not None:
+        record_buffer = recorder.stop_recording()
         logging.debug('Stopped recording (got %s byte(s))' % len(record_buffer))
 
         profile = request_to_profile(request, profiles_dirs)
@@ -632,15 +610,16 @@ def api_stop_recording():
 
 @app.route('/api/microphones', methods=['GET'])
 def api_microphones():
-    mics = {}
-    audio = pyaudio.PyAudio()
-    for i in range(audio.get_device_count()):
-        info = audio.get_device_info_by_index(i)
-        mics[i] = info['name']
+    profile = request_to_profile(request, profiles_dirs)
+    mics = get_audio_recorder(profile).get_microphones()
+    return jsonify(mics)
 
-    audio.terminate()
+def get_audio_recorder(profile):
+    system = profile.microphone.get('system', 'pyaudio')
+    if system == 'arecord':
+        return ARecordAudioRecorder()
 
-    return jsonify(mics)
+    return PyAudioRecorder()
 
 # -----------------------------------------------------------------------------
 

rhasspy/audio_recorder.py

@@ -0,0 +1,107 @@
+import os
+import re
+import subprocess
+import tempfile
+
+class AudioRecorder:
+    def start_recording(self, profile):
+        pass
+
+    def stop_recording(self):
+        return []
+
+    def get_microphones(self):
+        return {}
+
+# -----------------------------------------------------------------------------
+
+class PyAudioRecorder(AudioRecorder):
+    def __init__(self):
+        self.audio = None
+        self.mic = None
+        self.buffer = None
+
+    def start_recording(self, device_index=None):
+        import pyaudio
+
+        def stream_callback(data, frame_count, time_info, status):
+            if self.buffer is None:
+                self.buffer = data
+            else:
+                self.buffer += data
+
+            return (data, pyaudio.paContinue)
+
+        self.audio = pyaudio.PyAudio()
+        data_format = self.audio.get_format_from_width(2)  # 16-bit
+        self.mic = self.audio.open(format=data_format,
+                                   channels=1,
+                                   rate=16000,
+                                   input_device_index=device_index,
+                                   input=True,
+                                   stream_callback=stream_callback)
+
+        self.mic.start_stream()
+
+    def stop_recording(self):
+        self.mic.stop_stream()
+        self.audio.terminate()
+
+        return self.buffer
+
+    def get_microphones(self):
+        import pyaudio
+
+        mics = {}
+        audio = pyaudio.PyAudio()
+        for i in range(audio.get_device_count()):
+            info = audio.get_device_info_by_index(i)
+            mics[i] = info['name']
+
+        audio.terminate()
+
+        return mics
+
+# -----------------------------------------------------------------------------
+
+class ARecordAudioRecorder(AudioRecorder):
+    def start_recording(self, profile):
+        self.record_file = tempfile.NamedTemporaryFile(suffix='.wav', mode='wb+')
+        self.record_proc = subprocess.Popen(['arecord',
+                                             '-r', '16000',
+                                             '-f', 'S16_LE',
+                                             '-c', '1',
+                                             '-t', 'wav',
+                                             self.record_file.name],
+                                            close_fds=True)
+
+    def stop_recording(self):
+        self.record_proc.terminate()
+        self.record_file.seek(0)
+        data = open(self.record_file.name, 'rb').read()
+        try:
+            os.unlink(self.record_file.name)
+        except:
+            pass
+
+        return data
+
+    def get_microphones(self):
+        output = subprocess.check_output(['arecord', '-L'])\
+                           .decode().splitlines()
+
+        mics = {}
+        name, description = None, None
+
+        # Parse output of arecord -L
+        for line in output:
+            line = line.rstrip()
+            if re.match(r'^\s', line):
+                description = line.strip()
+            else:
+                if name is not None:
+                    mics[name] = description
+
+                name = line.strip()
+
+        return mics

rhasspy/config.json

@@ -1,7 +1,7 @@
 {
     "name": "Rhasspy Assistant",
     "slug": "rhasspy",
-    "version": "1.12",
+    "version": "1.13",
     "description": "Offline voice assistant for Home Assistant",
     "startup": "application",
     "boot": "auto",

rhasspy/dist/index.html

@@ -1 +1 @@
-<!DOCTYPE html><html lang=en><head><meta charset=utf-8><meta http-equiv=X-UA-Compatible content="IE=edge"><meta name=viewport content="width=device-width,initial-scale=1"><link rel=icon href=/img/favicon.png><link rel=stylesheet href=/css/bootstrap.min.css><link rel=stylesheet href=/css/fontawesome-all.min.css><link rel=stylesheet href=/css/main.css><title>Rhasspy Voice Assistant</title><link href=/css/app.acd79e7b.css rel=preload as=style><link href=/js/app.facf817a.js rel=preload as=script><link href=/js/chunk-vendors.fd9734c4.js rel=preload as=script><link href=/css/app.acd79e7b.css rel=stylesheet></head><body><noscript><strong>Please enable Javascript to continue.</strong></noscript><div id=app></div><script src=/js/jquery-3.3.1.slim.min.js></script><script src=/js/axios.min.js></script><script src=/js/vue-axios.min.js></script><script src=/js/bootstrap.min.js></script><script src=/js/chunk-vendors.fd9734c4.js></script><script src=/js/app.facf817a.js></script></body></html>
+<!DOCTYPE html><html lang=en><head><meta charset=utf-8><meta http-equiv=X-UA-Compatible content="IE=edge"><meta name=viewport content="width=device-width,initial-scale=1"><link rel=icon href=/img/favicon.png><link rel=stylesheet href=/css/bootstrap.min.css><link rel=stylesheet href=/css/fontawesome-all.min.css><link rel=stylesheet href=/css/main.css><title>Rhasspy Voice Assistant</title><link href=/css/app.778eaa10.css rel=preload as=style><link href=/js/app.75eca726.js rel=preload as=script><link href=/js/chunk-vendors.fd9734c4.js rel=preload as=script><link href=/css/app.778eaa10.css rel=stylesheet></head><body><noscript><strong>Please enable Javascript to continue.</strong></noscript><div id=app></div><script src=/js/jquery-3.3.1.slim.min.js></script><script src=/js/axios.min.js></script><script src=/js/vue-axios.min.js></script><script src=/js/bootstrap.min.js></script><script src=/js/chunk-vendors.fd9734c4.js></script><script src=/js/app.75eca726.js></script></body></html>

rhasspy/profiles.py

@@ -37,6 +37,7 @@ def load_profile(self):
         self.intent = self.json.get('intent', {})
         self.wake = self.json.get('wake', {})
         self.training = self.json.get('training', {})
+        self.microphone = self.json.get('microphone', {})
 
     def read_path(self, *path_parts):
         for profiles_dir in self.profiles_dirs:

rhasspy/profiles/defaults.json

@@ -20,6 +20,9 @@
         },
         "system": "fuzzywuzzy"
     },
+    "microphone": {
+        "system": "pyaudio"
+    },
     "rhasspy": {
         "default_profile": "en",
         "listen_on_start": false,