macos google assistant

google_assistant.py

"""Sample that implements gRPC client for Google Assistant API."""
from google.assistant.embedded.v1alpha1 import embedded_assistant_pb2
from google.rpc import code_pb2

import logging
import os.path

from googlesamples.assistant import (
    assistant_helpers,
    audio_helpers,
    auth_helpers,
    common_settings
)

def main():
    logging.basicConfig(level=logging.INFO)

    api_endpoint = 'embeddedassistant.googleapis.com'

    # Load credentials.
    try:
        creds = auth_helpers.load_credentials(
            "credentials.json", scopes=[common_settings.ASSISTANT_OAUTH_SCOPE]
        )
    except Exception as e:
        logging.error('Error loading credentials: %s', e)
        logging.error('Run auth_helpers to initialize new OAuth2 credentials.')
        return

    grpc_channel = auth_helpers.create_grpc_channel(api_endpoint, creds)
    logging.info('Connecting to %s', api_endpoint)
    assistant = embedded_assistant_pb2.EmbeddedAssistantStub(grpc_channel)

    audio_source = audio_helpers.SoundDeviceStream(
        sample_rate=16000,
        sample_width=2,
        block_size=6400,
        flush_size=25600
    )

    audio_sink = audio_helpers.SoundDeviceStream(
        sample_rate=16000,
        sample_width=2,
        block_size=6400,
        flush_size=25600
    )

    conversation_stream = audio_helpers.ConversationStream(
        source=audio_source,
        sink=audio_sink,
        iter_size=3200
    )
    
    conversation_state_bytes = None

    wait_for_user_trigger = False
    keep_running = True

    while keep_running:
        if wait_for_user_trigger:
            input('Press Enter to send a new request...')
        conversation_stream.start_recording()
        logging.info('Recording audio request.')

        def gen_converse_requests():
            converse_state = None
            if conversation_state_bytes:
                logging.debug('Sending converse_state: %s',
                              conversation_state_bytes)
                converse_state = embedded_assistant_pb2.ConverseState(
                    conversation_state=conversation_state_bytes,
                )
            config = embedded_assistant_pb2.ConverseConfig(
                audio_in_config=embedded_assistant_pb2.AudioInConfig(
                    encoding='LINEAR16',
                    sample_rate_hertz=16000
                ),
                audio_out_config=embedded_assistant_pb2.AudioOutConfig(
                    encoding='LINEAR16',
                    sample_rate_hertz=16000,
                    volume_percentage=50
                ),
                converse_state=converse_state
            )
            yield embedded_assistant_pb2.ConverseRequest(config=config)
            for data in conversation_stream:
                yield embedded_assistant_pb2.ConverseRequest(audio_in=data)

        def iter_converse_requests():
            for c in gen_converse_requests():
                assistant_helpers.log_converse_request_without_audio(c)
                yield c
            conversation_stream.start_playback()

        for resp in assistant.Converse(iter_converse_requests(), 185):
            assistant_helpers.log_converse_response_without_audio(resp)
            if resp.error.code != code_pb2.OK:
                logging.error('server error: %s', resp.error.message)
                break
            if resp.event_type == embedded_assistant_pb2.ConverseResponse.END_OF_UTTERANCE:
                logging.info('End of audio request detected')
                conversation_stream.stop_recording()
            if resp.result.spoken_request_text:
                logging.info('Transcript of user request: "%s".',
                             resp.result.spoken_request_text)
                logging.info('Playing assistant response.')
            if len(resp.audio_out.audio_data) > 0:
                conversation_stream.write(resp.audio_out.audio_data)
            if resp.result.spoken_response_text:
                logging.info(
                    'Transcript of TTS response '
                    '(only populated from IFTTT): "%s".',
                    resp.result.spoken_response_text)
            if resp.result.conversation_state:
                conversation_state_bytes = resp.result.conversation_state
            if resp.result.volume_percentage != 0:
                logging.info('Volume should be set to %s%%', resp.result.volume_percentage)
            if resp.result.microphone_mode == embedded_assistant_pb2.ConverseResult.DIALOG_FOLLOW_ON:
                wait_for_user_trigger = False
                logging.info('Expecting follow-on query from user.')
            elif resp.result.microphone_mode == embedded_assistant_pb2.ConverseResult.CLOSE_MICROPHONE:
                wait_for_user_trigger = True
                keep_running = False
        logging.info('Finished playing assistant response.')
        conversation_stream.stop_playback()

    conversation_stream.close()


if __name__ == '__main__':
    main()

google_assistant.sh

#!/bin/bash
source env/bin/activate
python3.5 google_assistant.py