Skip to content

Commit e34f45a

Browse files
jpvajdajpvajda
committed
adds readme info and example
1 parent 98095de commit e34f45a

File tree

2 files changed

+214
-1
lines changed

2 files changed

+214
-1
lines changed

‎README.md

Lines changed: 49 additions & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -249,8 +249,10 @@ options.agent.think.provider.model = "gpt-4o-mini"
249249
options.agent.think.prompt = "You are a helpful AI assistant."
250250
options.agent.listen.provider.type = "deepgram"
251251
options.agent.listen.provider.model = "nova-3"
252+
253+
# Option 1: Single TTS provider (backward compatible)
252254
options.agent.speak.provider.type = "deepgram"
253-
options.agent.speak.provider.model ="aura-2-thalia-en"
255+
options.agent.speak.provider.model = "aura-2-thalia-en"
254256

255257
options.greeting = "Hello, I'm your AI assistant."
256258

@@ -261,6 +263,52 @@ connection.start(options)
261263
connection.finish()
262264
```
263265

266+
### Multiple TTS Providers (Fallback Support)
267+
268+
For enhanced reliability, configure multiple TTS providers with automatic fallback:
269+
270+
```python
271+
from deepgram import (
272+
SettingsOptions,
273+
Speak,
274+
Endpoint,
275+
Header
276+
)
277+
278+
# Primary TTS provider
279+
primary_tts = Speak()
280+
primary_tts.provider.type = "deepgram"
281+
primary_tts.provider.model = "aura-2-zeus-en"
282+
283+
# Fallback TTS provider
284+
fallback_tts = Speak()
285+
fallback_tts.provider.type = "open_ai"
286+
fallback_tts.provider.model = "tts-1"
287+
fallback_tts.provider.voice = "shimmer"
288+
fallback_tts.endpoint = Endpoint()
289+
fallback_tts.endpoint.url = "https://api.openai.com/v1/audio/speech"
290+
fallback_tts.endpoint.headers = [
291+
Header(key="authorization", value="Bearer {{OPENAI_API_KEY}}")
292+
]
293+
294+
# Configure agent with fallback providers
295+
options = SettingsOptions()
296+
options.language = "en"
297+
options.agent.think.provider.type = "open_ai"
298+
options.agent.think.provider.model = "gpt-4o-mini"
299+
options.agent.think.prompt = "You are a helpful AI assistant."
300+
options.agent.listen.provider.type = "deepgram"
301+
options.agent.listen.provider.model = "nova-3"
302+
303+
# Option 2: Multiple TTS providers (with fallback)
304+
options.agent.speak = [primary_tts, fallback_tts]
305+
306+
options.greeting = "Hello, I'm your AI assistant with fallback TTS."
307+
308+
# Start the connection
309+
connection.start(options)
310+
```
311+
264312
This example demonstrates:
265313

266314
- Setting up a WebSocket connection

‎examples/agent/fallback_tts/main.py

Lines changed: 165 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,165 @@
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+
5+
"""
6+
Example demonstrating Voice Agent with multiple TTS providers for fallback support.
7+
8+
This example shows how to configure multiple TTS providers in an array format,
9+
allowing the agent to automatically fall back to alternative providers if the
10+
primary provider fails or is unavailable.
11+
"""
12+
13+
from deepgram.utils import verboselogs
14+
15+
from deepgram import (
16+
DeepgramClient,
17+
DeepgramClientOptions,
18+
AgentWebSocketEvents,
19+
SettingsOptions,
20+
Speak,
21+
Endpoint,
22+
Header,
23+
)
24+
25+
global warning_notice
26+
warning_notice = True
27+
28+
29+
def main():
30+
try:
31+
print("Starting Voice Agent with Fallback TTS Providers...")
32+
33+
config: DeepgramClientOptions = DeepgramClientOptions(
34+
options={
35+
"keepalive": "true",
36+
"microphone_record": "true",
37+
"speaker_playback": "true",
38+
},
39+
verbose=verboselogs.SPAM,
40+
)
41+
42+
deepgram: DeepgramClient = DeepgramClient("", config)
43+
dg_connection = deepgram.agent.websocket.v("1")
44+
45+
def on_open(self, open, **kwargs):
46+
print(f"\n\n{open}\n\n")
47+
48+
def on_binary_data(self, data, **kwargs):
49+
global warning_notice
50+
if warning_notice:
51+
print("Received binary data from TTS provider")
52+
print("Audio will be played automatically with speaker_playback=true")
53+
warning_notice = False
54+
55+
def on_welcome(self, welcome, **kwargs):
56+
print(f"\n\n{welcome}\n\n")
57+
58+
def on_settings_applied(self, settings_applied, **kwargs):
59+
print(f"\n\nSettings Applied - Multiple TTS providers configured:\n{settings_applied}\n\n")
60+
61+
def on_conversation_text(self, conversation_text, **kwargs):
62+
print(f"\n\n{conversation_text}\n\n")
63+
64+
def on_user_started_speaking(self, user_started_speaking, **kwargs):
65+
print(f"\n\n{user_started_speaking}\n\n")
66+
67+
def on_agent_thinking(self, agent_thinking, **kwargs):
68+
print(f"\n\n{agent_thinking}\n\n")
69+
70+
def on_agent_started_speaking(self, agent_started_speaking, **kwargs):
71+
print(f"\n\n{agent_started_speaking}\n\n")
72+
73+
def on_agent_audio_done(self, agent_audio_done, **kwargs):
74+
print(f"\n\n{agent_audio_done}\n\n")
75+
76+
def on_close(self, close, **kwargs):
77+
print(f"\n\n{close}\n\n")
78+
79+
def on_error(self, error, **kwargs):
80+
print(f"\n\nError (may trigger TTS fallback): {error}\n\n")
81+
82+
def on_unhandled(self, unhandled, **kwargs):
83+
print(f"\n\n{unhandled}\n\n")
84+
85+
# Register event handlers
86+
dg_connection.on(AgentWebSocketEvents.Open, on_open)
87+
dg_connection.on(AgentWebSocketEvents.AudioData, on_binary_data)
88+
dg_connection.on(AgentWebSocketEvents.Welcome, on_welcome)
89+
dg_connection.on(AgentWebSocketEvents.SettingsApplied, on_settings_applied)
90+
dg_connection.on(AgentWebSocketEvents.ConversationText, on_conversation_text)
91+
dg_connection.on(AgentWebSocketEvents.UserStartedSpeaking, on_user_started_speaking)
92+
dg_connection.on(AgentWebSocketEvents.AgentThinking, on_agent_thinking)
93+
dg_connection.on(AgentWebSocketEvents.AgentStartedSpeaking, on_agent_started_speaking)
94+
dg_connection.on(AgentWebSocketEvents.AgentAudioDone, on_agent_audio_done)
95+
dg_connection.on(AgentWebSocketEvents.Close, on_close)
96+
dg_connection.on(AgentWebSocketEvents.Error, on_error)
97+
dg_connection.on(AgentWebSocketEvents.Unhandled, on_unhandled)
98+
99+
# Configure TTS providers for fallback support
100+
print("Configuring TTS providers with fallback...")
101+
102+
# Primary TTS Provider: Deepgram Aura
103+
primary_tts = Speak()
104+
primary_tts.provider.type = "deepgram"
105+
primary_tts.provider.model = "aura-2-zeus-en"
106+
107+
# Fallback TTS Provider: OpenAI TTS
108+
fallback_tts = Speak()
109+
fallback_tts.provider.type = "open_ai"
110+
fallback_tts.provider.model = "tts-1"
111+
fallback_tts.provider.voice = "shimmer"
112+
113+
# Configure custom endpoint for OpenAI
114+
fallback_tts.endpoint = Endpoint()
115+
fallback_tts.endpoint.url = "https://api.openai.com/v1/audio/speech"
116+
fallback_tts.endpoint.headers = [
117+
Header(key="authorization", value="Bearer {{OPENAI_API_KEY}}")
118+
]
119+
120+
# Configure agent settings
121+
options: SettingsOptions = SettingsOptions()
122+
options.agent.think.provider.type = "open_ai"
123+
options.agent.think.provider.model = "gpt-4o-mini"
124+
options.agent.think.prompt = (
125+
"You are a helpful AI assistant with fallback TTS providers for reliable speech output. "
126+
"If one provider fails, another will automatically take over."
127+
)
128+
options.agent.listen.provider.type = "deepgram"
129+
options.agent.listen.provider.model = "nova-3"
130+
options.agent.listen.provider.keyterms = ["hello", "goodbye", "fallback"]
131+
132+
# Configure multiple TTS providers (array format)
133+
options.agent.speak = [primary_tts, fallback_tts]
134+
135+
options.agent.language = "en"
136+
options.greeting = (
137+
"Hello! I'm your AI assistant with fallback TTS providers. "
138+
"I can automatically switch between Deepgram and OpenAI for reliable voice output."
139+
)
140+
141+
print("TTS Providers configured:")
142+
print(f"1. Primary: {primary_tts.provider.type} - {primary_tts.provider.model}")
143+
print(f"2. Fallback: {fallback_tts.provider.type} - {fallback_tts.provider.model}")
144+
145+
# Start the connection
146+
if dg_connection.start(options) is False:
147+
print("Failed to start connection")
148+
return
149+
150+
print("\n\n=== Voice Agent with Fallback TTS Started ===")
151+
print("The agent is now running with primary and fallback TTS providers.")
152+
print("If the primary provider fails, it will automatically fall back to the secondary.")
153+
print("Press Enter to stop...\n\n")
154+
input()
155+
156+
# Close the connection
157+
dg_connection.finish()
158+
print("Finished - Fallback TTS example completed")
159+
160+
except Exception as e:
161+
print(f"An unexpected error occurred: {e}")
162+
163+
164+
if __name__ == "__main__":
165+
main()

0 commit comments

Comments
 (0)