|
| 1 | +# Copyright 2024 Deepgram SDK contributors. All Rights Reserved. |
| 2 | +# Use of this source code is governed by a MIT license that can be found in the LICENSE file. |
| 3 | +# SPDX-License-Identifier: MIT |
| 4 | +from signal import SIGINT, SIGTERM |
| 5 | +import asyncio |
| 6 | +import time |
| 7 | +from deepgram.utils import verboselogs |
| 8 | +from deepgram import ( |
| 9 | + DeepgramClient, |
| 10 | + DeepgramClientOptions, |
| 11 | + AgentWebSocketEvents, |
| 12 | + SettingsOptions, |
| 13 | +) |
| 14 | +TTS_TEXT = "Hello, this is a text to speech example using Deepgram." |
| 15 | +global warning_notice |
| 16 | +warning_notice = True |
| 17 | +async def main(): |
| 18 | + try: |
| 19 | + loop = asyncio.get_event_loop() |
| 20 | + for signal in (SIGTERM, SIGINT): |
| 21 | + loop.add_signal_handler( |
| 22 | + signal, |
| 23 | + lambda: asyncio.create_task(shutdown(signal, loop, dg_connection)), |
| 24 | + ) |
| 25 | + # example of setting up a client config. logging values: WARNING, VERBOSE, DEBUG, SPAM |
| 26 | + config: DeepgramClientOptions = DeepgramClientOptions( |
| 27 | + options={ |
| 28 | + "keepalive": "true", |
| 29 | + "microphone_record": "true", |
| 30 | + "speaker_playback": "true", |
| 31 | + }, |
| 32 | + # verbose=verboselogs.DEBUG, |
| 33 | + ) |
| 34 | + # Initialize Deepgram client - API key should be set in DEEPGRAM_API_KEY environment variable |
| 35 | + # For production testing, make sure your API key has proper permissions |
| 36 | + deepgram: DeepgramClient = DeepgramClient("", config) |
| 37 | + print("Initialized Deepgram client for production API testing") |
| 38 | + # Create a websocket connection to Deepgram |
| 39 | + dg_connection = deepgram.agent.asyncwebsocket.v("1") |
| 40 | + async def on_open(self, open, **kwargs): |
| 41 | + print(f"\n\n{open}\n\n") |
| 42 | + async def on_binary_data(self, data, **kwargs): |
| 43 | + global warning_notice |
| 44 | + if warning_notice: |
| 45 | + print("Received binary data") |
| 46 | + print("You can do something with the binary data here") |
| 47 | + print("OR") |
| 48 | + print( |
| 49 | + "If you want to simply play the audio, set speaker_playback to true in the options for DeepgramClientOptions" |
| 50 | + ) |
| 51 | + warning_notice = False |
| 52 | + async def on_welcome(self, welcome, **kwargs): |
| 53 | + print(f"\n\n{welcome}\n\n") |
| 54 | + async def on_settings_applied(self, settings_applied, **kwargs): |
| 55 | + print(f"\n\n{settings_applied}\n\n") |
| 56 | + async def on_conversation_text(self, conversation_text, **kwargs): |
| 57 | + print(f"\n\n{conversation_text}\n\n") |
| 58 | + async def on_user_started_speaking(self, user_started_speaking, **kwargs): |
| 59 | + print(f"\n\n{user_started_speaking}\n\n") |
| 60 | + async def on_agent_thinking(self, agent_thinking, **kwargs): |
| 61 | + print(f"\n\n{agent_thinking}\n\n") |
| 62 | + async def on_agent_started_speaking(self, agent_started_speaking, **kwargs): |
| 63 | + print(f"\n\n{agent_started_speaking}\n\n") |
| 64 | + async def on_agent_audio_done(self, agent_audio_done, **kwargs): |
| 65 | + print(f"\n\n{agent_audio_done}\n\n") |
| 66 | + async def on_close(self, close, **kwargs): |
| 67 | + print(f"\n\n{close}\n\n") |
| 68 | + async def on_error(self, error, **kwargs): |
| 69 | + print(f"\n\n{error}\n\n") |
| 70 | + async def on_unhandled(self, unhandled, **kwargs): |
| 71 | + print(f"\n\n{unhandled}\n\n") |
| 72 | + dg_connection.on(AgentWebSocketEvents.Open, on_open) |
| 73 | + dg_connection.on(AgentWebSocketEvents.AudioData, on_binary_data) |
| 74 | + dg_connection.on(AgentWebSocketEvents.Welcome, on_welcome) |
| 75 | + dg_connection.on(AgentWebSocketEvents.SettingsApplied, on_settings_applied) |
| 76 | + dg_connection.on(AgentWebSocketEvents.ConversationText, on_conversation_text) |
| 77 | + dg_connection.on( |
| 78 | + AgentWebSocketEvents.UserStartedSpeaking, on_user_started_speaking |
| 79 | + ) |
| 80 | + dg_connection.on(AgentWebSocketEvents.AgentThinking, on_agent_thinking) |
| 81 | + dg_connection.on( |
| 82 | + AgentWebSocketEvents.AgentStartedSpeaking, on_agent_started_speaking |
| 83 | + ) |
| 84 | + dg_connection.on(AgentWebSocketEvents.AgentAudioDone, on_agent_audio_done) |
| 85 | + dg_connection.on(AgentWebSocketEvents.Close, on_close) |
| 86 | + dg_connection.on(AgentWebSocketEvents.Error, on_error) |
| 87 | + dg_connection.on(AgentWebSocketEvents.Unhandled, on_unhandled) |
| 88 | + # connect to websocket |
| 89 | + options = SettingsOptions() |
| 90 | + options.agent.think.provider.type = "open_ai" |
| 91 | + options.agent.think.provider.model = "gpt-4o-mini" |
| 92 | + options.agent.think.prompt = "You are a helpful AI assistant." |
| 93 | + options.greeting = "Hello, this is a text to speech example using Deepgram." |
| 94 | + options.agent.listen.provider.keyterms = ["hello", "goodbye"] |
| 95 | + options.agent.listen.provider.model = "nova-3" |
| 96 | + options.agent.listen.provider.type = "deepgram" |
| 97 | + options.agent.speak.provider.type = "deepgram" |
| 98 | + options.agent.speak.provider.model = "aura-2-thalia-en" |
| 99 | + options.agent.language = "en" |
| 100 | + # Add tags for production testing |
| 101 | + options.tags = ["production-test", "sdk-example", "agent-websocket", "tags-validation"] |
| 102 | + print(f"Using tags: {options.tags}") |
| 103 | + # Print the full options being sent |
| 104 | + print("Options being sent to API:") |
| 105 | + print(options.to_json()) |
| 106 | + print("\n\n✅ Connection established with tags!") |
| 107 | + print(f"✅ Tags being used: {options.tags}") |
| 108 | + print("\n🎤 You can now speak into your microphone...") |
| 109 | + print("The agent will respond using the production API with tags.") |
| 110 | + print("Press Ctrl+C to stop.\n\n") |
| 111 | + if await dg_connection.start(options) is False: |
| 112 | + print("Failed to start connection") |
| 113 | + return |
| 114 | + # wait until cancelled |
| 115 | + try: |
| 116 | + while True: |
| 117 | + await asyncio.sleep(1) |
| 118 | + except asyncio.CancelledError: |
| 119 | + # This block will be executed when the shutdown coroutine cancels all tasks |
| 120 | + pass |
| 121 | + finally: |
| 122 | + await dg_connection.finish() |
| 123 | + print("Finished") |
| 124 | + except ValueError as e: |
| 125 | + print(f"Invalid value encountered: {e}") |
| 126 | + except Exception as e: |
| 127 | + print(f"An unexpected error occurred: {e}") |
| 128 | +async def shutdown(signal, loop, dg_connection): |
| 129 | + print(f"Received exit signal {signal.name}...") |
| 130 | + await dg_connection.finish() |
| 131 | + tasks = [t for t in asyncio.all_tasks() if t is not asyncio.current_task()] |
| 132 | + [task.cancel() for task in tasks] |
| 133 | + print(f"Cancelling {len(tasks)} outstanding tasks") |
| 134 | + await asyncio.gather(*tasks, return_exceptions=True) |
| 135 | + loop.stop() |
| 136 | + print("Shutdown complete.") |
| 137 | +asyncio.run(main()) |
0 commit comments