update gemini voice to use gemini-gradio
Browse files- app_gemini_voice.py +12 -188
- requirements.txt +1 -1
app_gemini_voice.py
CHANGED
|
@@ -1,194 +1,18 @@
|
|
| 1 |
-
import base64
|
| 2 |
-
import json
|
| 3 |
import os
|
| 4 |
|
| 5 |
-
import
|
| 6 |
-
import numpy as np
|
| 7 |
-
import websockets.sync.client
|
| 8 |
-
from dotenv import load_dotenv
|
| 9 |
-
from gradio_webrtc import StreamHandler, WebRTC, get_twilio_turn_credentials
|
| 10 |
|
|
|
|
| 11 |
|
| 12 |
-
|
| 13 |
-
|
| 14 |
-
|
| 15 |
-
|
| 16 |
-
|
| 17 |
-
|
| 18 |
-
|
| 19 |
-
|
| 20 |
-
|
| 21 |
-
api_key = os.getenv("GOOGLE_API_KEY")
|
| 22 |
-
if not api_key:
|
| 23 |
-
raise ValueError("GOOGLE_API_KEY not found in environment variables. Please set it in your .env file.")
|
| 24 |
-
return api_key
|
| 25 |
-
|
| 26 |
-
|
| 27 |
-
class AudioProcessor:
|
| 28 |
-
@staticmethod
|
| 29 |
-
def encode_audio(data, sample_rate):
|
| 30 |
-
encoded = base64.b64encode(data.tobytes()).decode("UTF-8")
|
| 31 |
-
return {
|
| 32 |
-
"realtimeInput": {
|
| 33 |
-
"mediaChunks": [
|
| 34 |
-
{
|
| 35 |
-
"mimeType": f"audio/pcm;rate={sample_rate}",
|
| 36 |
-
"data": encoded,
|
| 37 |
-
}
|
| 38 |
-
],
|
| 39 |
-
},
|
| 40 |
-
}
|
| 41 |
-
|
| 42 |
-
@staticmethod
|
| 43 |
-
def process_audio_response(data):
|
| 44 |
-
audio_data = base64.b64decode(data)
|
| 45 |
-
return np.frombuffer(audio_data, dtype=np.int16)
|
| 46 |
-
|
| 47 |
-
|
| 48 |
-
class GeminiHandler(StreamHandler):
|
| 49 |
-
def __init__(self, expected_layout="mono", output_sample_rate=24000, output_frame_size=480) -> None:
|
| 50 |
-
super().__init__(expected_layout, output_sample_rate, output_frame_size, input_sample_rate=24000)
|
| 51 |
-
self.config = GeminiConfig()
|
| 52 |
-
self.ws = None
|
| 53 |
-
self.all_output_data = None
|
| 54 |
-
self.audio_processor = AudioProcessor()
|
| 55 |
-
|
| 56 |
-
def copy(self):
|
| 57 |
-
return GeminiHandler(
|
| 58 |
-
expected_layout=self.expected_layout,
|
| 59 |
-
output_sample_rate=self.output_sample_rate,
|
| 60 |
-
output_frame_size=self.output_frame_size,
|
| 61 |
-
)
|
| 62 |
-
|
| 63 |
-
def _initialize_websocket(self):
|
| 64 |
-
try:
|
| 65 |
-
self.ws = websockets.sync.client.connect(self.config.ws_url, timeout=30)
|
| 66 |
-
initial_request = {
|
| 67 |
-
"setup": {
|
| 68 |
-
"model": self.config.model,
|
| 69 |
-
}
|
| 70 |
-
}
|
| 71 |
-
self.ws.send(json.dumps(initial_request))
|
| 72 |
-
setup_response = json.loads(self.ws.recv())
|
| 73 |
-
print(f"Setup response: {setup_response}")
|
| 74 |
-
except websockets.exceptions.WebSocketException as e:
|
| 75 |
-
print(f"WebSocket connection failed: {str(e)}")
|
| 76 |
-
self.ws = None
|
| 77 |
-
except Exception as e:
|
| 78 |
-
print(f"Setup failed: {str(e)}")
|
| 79 |
-
self.ws = None
|
| 80 |
-
|
| 81 |
-
def receive(self, frame: tuple[int, np.ndarray]) -> None:
|
| 82 |
-
try:
|
| 83 |
-
if not self.ws:
|
| 84 |
-
self._initialize_websocket()
|
| 85 |
-
|
| 86 |
-
_, array = frame
|
| 87 |
-
array = array.squeeze()
|
| 88 |
-
audio_message = self.audio_processor.encode_audio(array, self.output_sample_rate)
|
| 89 |
-
self.ws.send(json.dumps(audio_message)) # type: ignore
|
| 90 |
-
except Exception as e:
|
| 91 |
-
print(f"Error in receive: {str(e)}")
|
| 92 |
-
if self.ws:
|
| 93 |
-
self.ws.close()
|
| 94 |
-
self.ws = None
|
| 95 |
-
|
| 96 |
-
def _process_server_content(self, content):
|
| 97 |
-
for part in content.get("parts", []):
|
| 98 |
-
data = part.get("inlineData", {}).get("data", "")
|
| 99 |
-
if data:
|
| 100 |
-
audio_array = self.audio_processor.process_audio_response(data)
|
| 101 |
-
if self.all_output_data is None:
|
| 102 |
-
self.all_output_data = audio_array
|
| 103 |
-
else:
|
| 104 |
-
self.all_output_data = np.concatenate((self.all_output_data, audio_array))
|
| 105 |
-
|
| 106 |
-
while self.all_output_data.shape[-1] >= self.output_frame_size:
|
| 107 |
-
yield (self.output_sample_rate, self.all_output_data[: self.output_frame_size].reshape(1, -1))
|
| 108 |
-
self.all_output_data = self.all_output_data[self.output_frame_size :]
|
| 109 |
-
|
| 110 |
-
def generator(self):
|
| 111 |
-
while True:
|
| 112 |
-
if not self.ws:
|
| 113 |
-
print("WebSocket not connected")
|
| 114 |
-
yield None
|
| 115 |
-
continue
|
| 116 |
-
|
| 117 |
-
try:
|
| 118 |
-
message = self.ws.recv(timeout=5)
|
| 119 |
-
msg = json.loads(message)
|
| 120 |
-
|
| 121 |
-
if "serverContent" in msg:
|
| 122 |
-
content = msg["serverContent"].get("modelTurn", {})
|
| 123 |
-
yield from self._process_server_content(content)
|
| 124 |
-
except TimeoutError:
|
| 125 |
-
print("Timeout waiting for server response")
|
| 126 |
-
yield None
|
| 127 |
-
except Exception as e:
|
| 128 |
-
print(f"Error in generator: {str(e)}")
|
| 129 |
-
yield None
|
| 130 |
-
|
| 131 |
-
def emit(self) -> tuple[int, np.ndarray] | None:
|
| 132 |
-
if not self.ws:
|
| 133 |
-
return None
|
| 134 |
-
if not hasattr(self, "_generator"):
|
| 135 |
-
self._generator = self.generator()
|
| 136 |
-
try:
|
| 137 |
-
return next(self._generator)
|
| 138 |
-
except StopIteration:
|
| 139 |
-
self.reset()
|
| 140 |
-
return None
|
| 141 |
-
|
| 142 |
-
def reset(self) -> None:
|
| 143 |
-
if hasattr(self, "_generator"):
|
| 144 |
-
delattr(self, "_generator")
|
| 145 |
-
self.all_output_data = None
|
| 146 |
-
|
| 147 |
-
def shutdown(self) -> None:
|
| 148 |
-
if self.ws:
|
| 149 |
-
self.ws.close()
|
| 150 |
-
|
| 151 |
-
def check_connection(self):
|
| 152 |
-
try:
|
| 153 |
-
if not self.ws or self.ws.closed:
|
| 154 |
-
self._initialize_websocket()
|
| 155 |
-
return True
|
| 156 |
-
except Exception as e:
|
| 157 |
-
print(f"Connection check failed: {str(e)}")
|
| 158 |
-
return False
|
| 159 |
-
|
| 160 |
-
|
| 161 |
-
class GeminiVoiceChat:
|
| 162 |
-
def __init__(self):
|
| 163 |
-
load_dotenv()
|
| 164 |
-
self.demo = self._create_interface()
|
| 165 |
-
|
| 166 |
-
def _create_interface(self):
|
| 167 |
-
with gr.Blocks() as demo:
|
| 168 |
-
gr.HTML(
|
| 169 |
-
"""
|
| 170 |
-
<div style='text-align: center'>
|
| 171 |
-
<h1>Gemini 2.0 Voice Chat</h1>
|
| 172 |
-
<p>Speak with Gemini using real-time audio streaming</p>
|
| 173 |
-
</div>
|
| 174 |
-
"""
|
| 175 |
-
)
|
| 176 |
-
|
| 177 |
-
webrtc = WebRTC(
|
| 178 |
-
label="Conversation",
|
| 179 |
-
modality="audio",
|
| 180 |
-
mode="send-receive",
|
| 181 |
-
rtc_configuration=get_twilio_turn_credentials(),
|
| 182 |
-
)
|
| 183 |
-
|
| 184 |
-
webrtc.stream(GeminiHandler(), inputs=[webrtc], outputs=[webrtc], time_limit=90, concurrency_limit=10)
|
| 185 |
-
return demo
|
| 186 |
-
|
| 187 |
-
def launch(self):
|
| 188 |
-
self.demo.launch()
|
| 189 |
-
|
| 190 |
-
|
| 191 |
-
demo = GeminiVoiceChat().demo
|
| 192 |
|
| 193 |
if __name__ == "__main__":
|
| 194 |
-
demo.launch(
|
|
|
|
|
|
|
|
|
|
| 1 |
import os
|
| 2 |
|
| 3 |
+
import gemini_gradio
|
|
|
|
|
|
|
|
|
|
|
|
|
| 4 |
|
| 5 |
+
from utils import get_app
|
| 6 |
|
| 7 |
+
demo = get_app(
|
| 8 |
+
models=[
|
| 9 |
+
"gemini-2.0-flash-exp",
|
| 10 |
+
],
|
| 11 |
+
default_model="gemini-2.0-flash-exp",
|
| 12 |
+
src=gemini_gradio.registry,
|
| 13 |
+
accept_token=not os.getenv("GEMINI_API_KEY"),
|
| 14 |
+
enable_voice=True,
|
| 15 |
+
)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 16 |
|
| 17 |
if __name__ == "__main__":
|
| 18 |
+
demo.launch()
|
requirements.txt
CHANGED
|
@@ -113,7 +113,7 @@ fsspec==2024.10.0
|
|
| 113 |
# via
|
| 114 |
# gradio-client
|
| 115 |
# huggingface-hub
|
| 116 |
-
gemini-gradio==0.0.
|
| 117 |
# via anychat (pyproject.toml)
|
| 118 |
google-ai-generativelanguage==0.6.10
|
| 119 |
# via google-generativeai
|
|
|
|
| 113 |
# via
|
| 114 |
# gradio-client
|
| 115 |
# huggingface-hub
|
| 116 |
+
gemini-gradio==0.0.3
|
| 117 |
# via anychat (pyproject.toml)
|
| 118 |
google-ai-generativelanguage==0.6.10
|
| 119 |
# via google-generativeai
|