Voicekit - a problem about bluetooth

HACKING.md

@@ -1,5 +1,9 @@
 # Setting up the image
 
+**Deprecated:** The newest instructions are in the
+[aiyprojects
+branch](https://github.com/google/aiyprojects-raspbian/blob/aiyprojects/HACKING.md).
+
 We recommend using [the images](https://aiyprojects.withgoogle.com/voice) we
 provide. Those images are based on [Raspbian](https://www.raspberrypi.org/downloads/raspbian/),
 with a few customizations and are tested on the Raspberry Pi 3. If you prefer

README.md

@@ -1,4 +1,4 @@
-<img src="https://aiyprojects.withgoogle.com/static/images/icons/aiy-square-logo.svg" width="40%">
+<img src="https://aiyprojects.withgoogle.com/static/images/icons/aiy-circular-logo.svg" width="40%">
 
 This repository contains an easy-to-use API for the AIY Voice Kit.
 You can use it to create voice commands with simple while loops - have a look at the [demos](https://github.com/google/aiyprojects-raspbian/tree/voicekit/src).
@@ -9,8 +9,15 @@ If you're using Raspbian instead of Google's provided image, read
 [HACKING.md](HACKING.md) for information on getting started.
 
 For returning users:
-The old voice-recognizer demo remains in the [master branch](https://github.com/google/aiyprojects-raspbian/tree/master) of this project.
-The new code is in the `voicekit` branch, and is included in images starting with aiyprojects-2017-09-11.img.
+The newest code is in the
+[aiyprojects
+branch](https://github.com/google/aiyprojects-raspbian/tree/aiyprojects), which
+supports all AIY kits.
+The `voicekit` branch is deprecated, and users should switch to the
+`aiyprojects` branch.
+The original, deprecated voice-recognizer demo remains in the [master
+branch](https://github.com/google/aiyprojects-raspbian/tree/master) of this
+project.
 
 # Support
 

requirements.txt

@@ -1,3 +1,3 @@
-google-assistant-grpc==0.0.2
-grpc-google-cloud-speech-v1beta1==0.14.0
-google-auth-oauthlib==0.1.0
+google-assistant-grpc==0.1.0
+google-cloud-speech==0.30.0
+google-auth-oauthlib==0.2.0

scripts/install-deps.sh

@@ -33,7 +33,7 @@ cd "${scripts_dir}/.."
 virtualenv --system-site-packages -p python3 env
 env/bin/pip install -r requirements.txt
 
-# The google-assistant-library is only available on ARMv7.
-if [[ "$(uname -m)" == "armv7l" ]] ; then
-  env/bin/pip install google-assistant-library==0.0.3
+# The google-assistant-library is only available on some platforms.
+if [[ "$(uname -m)" == "armv7l" || "$(uname -m)" == "x86_64" ]] ; then
+  env/bin/pip install google-assistant-library==0.1.0
 fi

src/aiy/_apis/_speech.py

@@ -18,16 +18,28 @@
 import collections
 import logging
 import os
+import sys
 import tempfile
 import wave
 
 import google.auth
 import google.auth.exceptions
 import google.auth.transport.grpc
 import google.auth.transport.requests
-from google.cloud.grpc.speech.v1beta1 import cloud_speech_pb2 as cloud_speech
+try:
+    from google.cloud import speech
+    from google.cloud.speech import enums
+    from google.cloud.speech import types
+except ImportError:
+    print("Failed to import google.cloud.speech. Try:")
+    print("    env/bin/pip install -r requirements.txt")
+    sys.exit(1)
+
 from google.rpc import code_pb2 as error_code
-from google.assistant.embedded.v1alpha1 import embedded_assistant_pb2
+from google.assistant.embedded.v1alpha1 import (
+    embedded_assistant_pb2,
+    embedded_assistant_pb2_grpc,
+)
 import grpc
 from six.moves import queue
 
@@ -135,7 +147,7 @@ def _get_speech_context(self):
         """Return a SpeechContext instance to bias recognition towards certain
         phrases.
         """
-        return cloud_speech.SpeechContext(
+        return types.SpeechContext(
             phrases=self._phrases,
         )
 
@@ -289,53 +301,51 @@ def __init__(self, credentials_file):
 
         self.language_code = aiy.i18n.get_language_code()
 
-        if not hasattr(cloud_speech, 'StreamingRecognizeRequest'):
-            raise ValueError("cloud_speech_pb2.py doesn't have StreamingRecognizeRequest.")
-
         self._transcript = None
 
     def reset(self):
         super().reset()
         self._transcript = None
 
     def _make_service(self, channel):
-        return cloud_speech.SpeechStub(channel)
+        return speech.SpeechClient()
 
     def _create_config_request(self):
-        recognition_config = cloud_speech.RecognitionConfig(
-            # There are a bunch of config options you can specify. See
-            # https://goo.gl/KPZn97 for the full list.
-            encoding='LINEAR16',  # raw 16-bit signed LE samples
-            sample_rate=AUDIO_SAMPLE_RATE_HZ,
+        recognition_config = types.RecognitionConfig(
+            encoding=enums.RecognitionConfig.AudioEncoding.LINEAR16,
+            sample_rate_hertz=AUDIO_SAMPLE_RATE_HZ,
             # For a list of supported languages see:
             # https://cloud.google.com/speech/docs/languages.
             language_code=self.language_code,  # a BCP-47 language tag
-            speech_context=self._get_speech_context(),
+            speech_contexts=[self._get_speech_context()],
         )
-        streaming_config = cloud_speech.StreamingRecognitionConfig(
+        streaming_config = types.StreamingRecognitionConfig(
             config=recognition_config,
-            single_utterance=True,  # TODO(rodrigoq): find a way to handle pauses
+            single_utterance=True,
         )
 
-        return cloud_speech.StreamingRecognizeRequest(
-            streaming_config=streaming_config)
+        # TODO(rodrigoq): we're actually returning a Config, not a Request, as
+        # the v1 API takes the Config and wraps it up in a Request, but we still
+        # want to share code with the Assistant API. Can we clean this up?
+        return streaming_config
 
     def _create_audio_request(self, data):
-        return cloud_speech.StreamingRecognizeRequest(audio_content=data)
+        return types.StreamingRecognizeRequest(audio_content=data)
 
-    def _create_response_stream(self, service, request_stream, deadline):
-        return service.StreamingRecognize(request_stream, deadline)
+    def _create_response_stream(self, client, request_stream, deadline):
+        config = next(request_stream)
+        return client.streaming_recognize(config, request_stream)
 
     def _stop_sending_audio(self, resp):
         """Check the endpointer type to see if an utterance has ended."""
 
-        if resp.endpointer_type:
-            endpointer_type = cloud_speech.StreamingRecognizeResponse.EndpointerType.Name(
-                resp.endpointer_type)
-            logger.info('endpointer_type: %s', endpointer_type)
+        if resp.speech_event_type:
+            speech_event_type = types.StreamingRecognizeResponse.SpeechEventType.Name(
+                resp.speech_event_type)
+            logger.info('endpointer_type: %s', speech_event_type)
 
-        END_OF_AUDIO = cloud_speech.StreamingRecognizeResponse.EndpointerType.Value('END_OF_AUDIO')
-        return resp.endpointer_type == END_OF_AUDIO
+        END_OF_SINGLE_UTTERANCE = types.StreamingRecognizeResponse.SpeechEventType.Value('END_OF_SINGLE_UTTERANCE')
+        return resp.speech_event_type == END_OF_SINGLE_UTTERANCE
 
     def _handle_response(self, resp):
         """Store the last transcript we received."""
@@ -367,7 +377,7 @@ def reset(self):
         self._transcript = None
 
     def _make_service(self, channel):
-        return embedded_assistant_pb2.EmbeddedAssistantStub(channel)
+        return embedded_assistant_pb2_grpc.EmbeddedAssistantStub(channel)
 
     def _create_config_request(self):
         audio_in_config = embedded_assistant_pb2.AudioInConfig(

src/aiy/assistant/device_helpers.py

@@ -0,0 +1,118 @@
+#!/usr/bin/env python3
+# Copyright 2017 Google Inc.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+"""Device registration helpers for the Google Assistant API."""
+
+import json
+import os
+import uuid
+
+import google.auth.transport.requests
+
+import aiy.assistant.auth_helpers
+
+
+_DEVICE_MODEL = "voice-kit"
+_DEVICE_MANUFACTURER = "AIY Projects"
+_DEVICE_NAME = "Voice Kit"
+_DEVICE_TYPE = "action.devices.types.LIGHT"
+
+_DEVICE_ID_FILE = os.path.join(
+        aiy.assistant.auth_helpers._VR_CACHE_DIR, 'device_id.json')
+
+
+def _get_project_id():
+    with open(aiy.assistant.auth_helpers._ASSISTANT_CREDENTIALS_FILE) as f:
+        client_secrets_data = json.load(f)
+        return client_secrets_data["installed"]["project_id"]
+
+
+def _get_api_url(*args):
+    return "/".join(
+            ("https://embeddedassistant.googleapis.com/v1alpha2/projects",) + args)
+
+def _load_ids(id_path):
+    with open(id_path, 'r') as f:
+        id_data = json.load(f)
+    return id_data["model_id"], id_data["device_id"]
+
+
+def _save_ids(id_path, model_id, device_id):
+    if not os.path.exists(os.path.dirname(id_path)):
+        os.makedirs(os.path.dirname(id_path))
+
+    id_data = {
+            "model_id": model_id,
+            "device_id": device_id,
+    }
+    with open(id_path, 'w') as f:
+        json.dump(id_data, f)
+
+
+def _get_model_id(credentials, session, project_id):
+    model_id = "%s-%s" % (project_id, _DEVICE_MODEL)
+    payload = {
+            "device_model_id": model_id,
+            "project_id": project_id,
+            "device_type": _DEVICE_TYPE,
+            "manifest": {
+                    "manufacturer": _DEVICE_MANUFACTURER,
+                    "product_name": _DEVICE_NAME,
+            },
+    }
+    r = session.post(_get_api_url(project_id, "deviceModels"),
+                     data=json.dumps(payload))
+    # Ignore 409, which means we've already created the model ID.
+    if r.status_code != 409:
+        r.raise_for_status()
+    return model_id
+
+
+def get_ids(credentials, model_id=None):
+    """get_ids gets a Device ID for use with the Google Assistant SDK.
+
+    It optionally also gets a Device Model ID if one is not given. The IDs are
+    cached on disk so that a device keeps a consistent ID.
+
+    Returns:
+        a tuple: (model_id, device_id)
+    """
+
+    if os.path.exists(_DEVICE_ID_FILE):
+        return _load_ids(_DEVICE_ID_FILE)
+
+    session = google.auth.transport.requests.AuthorizedSession(credentials)
+    project_id = _get_project_id()
+    model_id = model_id or _get_model_id(credentials, session, project_id)
+
+    device_id = "%s-%s" % (model_id, uuid.uuid4())
+    # We can hardcode client_type as SDK_SERVICE, because the Assistant Library
+    # creates its own device_id.
+    payload = {
+            "id": device_id,
+            "model_id": model_id,
+            "client_type": "SDK_SERVICE",
+    }
+    r = session.post(_get_api_url(project_id, "devices"),
+                     data=json.dumps(payload))
+    r.raise_for_status()
+
+    _save_ids(_DEVICE_ID_FILE, model_id, device_id)
+    return model_id, device_id
+
+
+if __name__ == "__main__":
+    credentials = aiy.assistant.auth_helpers.get_assistant_credentials()
+    print("ids:", get_ids(credentials))

src/aiy/cloudspeech.py

@@ -36,18 +36,30 @@ def __init__(self, credentials_file):
         self._recorder = aiy.audio.get_recorder()
         self._hotwords = []
 
-    def recognize(self):
+    def recognize(self, immediate=False):
         """Recognizes the user's speech and transcript it into text.
 
         This function listens to the user's speech via the VoiceHat speaker. Then it
         contacts Google CloudSpeech APIs and returns a textual transcript if possible.
         If hotword list is populated this method will only respond if hotword is said.
+
+        Args:
+            immediate: ignore the hotword list, even if it has been populated
+                may be used to create a conversational experience, for example:
+
+                text = recognizer.recognize()
+                if 'call a friend' in text:
+                    aiy.audio.say('OK, which one?')
+                    friend = recognizer.recognize(immediate=True)
+                    make_a_call(friend)
         """
         self._request.reset()
         self._request.set_endpointer_cb(self._endpointer_callback)
         self._recorder.add_processor(self._request)
         text = self._request.do_request().transcript
-        if self._hotwords and text:
+        if immediate:
+            return text
+        elif self._hotwords and text:
             text = text.lower()
             loc_min = len(text)
             hotword_found = ''

src/assistant_library_demo.py

@@ -28,6 +28,7 @@
 import sys
 
 import aiy.assistant.auth_helpers
+import aiy.assistant.device_helpers
 import aiy.voicehat
 from google.assistant.library import Assistant
 from google.assistant.library.event import EventType
@@ -60,7 +61,8 @@ def process_event(event):
 
 def main():
     credentials = aiy.assistant.auth_helpers.get_assistant_credentials()
-    with Assistant(credentials) as assistant:
+    model_id, device_id = aiy.assistant.device_helpers.get_ids(credentials)
+    with Assistant(credentials, model_id) as assistant:
         for event in assistant.start():
             process_event(event)
 

src/assistant_library_with_button_demo.py

@@ -29,6 +29,7 @@
 import threading
 
 import aiy.assistant.auth_helpers
+import aiy.assistant.device_helpers
 import aiy.voicehat
 from google.assistant.library import Assistant
 from google.assistant.library.event import EventType
@@ -61,7 +62,8 @@ def start(self):
 
     def _run_task(self):
         credentials = aiy.assistant.auth_helpers.get_assistant_credentials()
-        with Assistant(credentials) as assistant:
+        model_id, device_id = aiy.assistant.device_helpers.get_ids(credentials)
+        with Assistant(credentials, model_id) as assistant:
             self._assistant = assistant
             for event in assistant.start():
                 self._process_event(event)

src/assistant_library_with_local_commands_demo.py

@@ -29,6 +29,7 @@
 import sys
 
 import aiy.assistant.auth_helpers
+import aiy.assistant.device_helpers
 import aiy.audio
 import aiy.voicehat
 from google.assistant.library import Assistant
@@ -90,7 +91,8 @@ def process_event(assistant, event):
 
 def main():
     credentials = aiy.assistant.auth_helpers.get_assistant_credentials()
-    with Assistant(credentials) as assistant:
+    model_id, device_id = aiy.assistant.device_helpers.get_ids(credentials)
+    with Assistant(credentials, model_id) as assistant:
         for event in assistant.start():
             process_event(assistant, event)