Spaces:
Runtime error
Runtime error
Update app.py
Browse files
app.py
CHANGED
|
@@ -3,8 +3,6 @@ import numpy as np
|
|
| 3 |
import io
|
| 4 |
from pydub import AudioSegment
|
| 5 |
import tempfile
|
| 6 |
-
import os
|
| 7 |
-
import base64
|
| 8 |
import openai
|
| 9 |
import time
|
| 10 |
from dataclasses import dataclass, field
|
|
@@ -14,11 +12,11 @@ from threading import Lock
|
|
| 14 |
class AppState:
|
| 15 |
stream: np.ndarray | None = None
|
| 16 |
sampling_rate: int = 0
|
| 17 |
-
|
| 18 |
-
last_speech: float = 0
|
| 19 |
conversation: list = field(default_factory=list)
|
| 20 |
client: openai.OpenAI = None
|
| 21 |
output_format: str = "mp3"
|
|
|
|
| 22 |
|
| 23 |
# Global lock for thread safety
|
| 24 |
state_lock = Lock()
|
|
@@ -29,27 +27,36 @@ def create_client(api_key):
|
|
| 29 |
api_key=api_key
|
| 30 |
)
|
| 31 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 32 |
def process_audio(audio: tuple, state: AppState):
|
| 33 |
if state.stream is None:
|
| 34 |
state.stream = audio[1]
|
| 35 |
state.sampling_rate = audio[0]
|
| 36 |
-
state.last_speech = time.time()
|
| 37 |
else:
|
| 38 |
state.stream = np.concatenate((state.stream, audio[1]))
|
| 39 |
|
| 40 |
-
|
| 41 |
-
|
| 42 |
-
if np.max(np.abs(audio[1])) > 0.1: # Adjust this threshold as needed
|
| 43 |
-
state.last_speech = current_time
|
| 44 |
-
state.pause_start = None
|
| 45 |
-
elif state.pause_start is None:
|
| 46 |
-
state.pause_start = current_time
|
| 47 |
|
| 48 |
-
|
| 49 |
-
if state.pause_start and (current_time - state.pause_start > 2.0): # 2 seconds of silence
|
| 50 |
return gr.Audio(recording=False), state
|
| 51 |
-
|
| 52 |
-
|
| 53 |
|
| 54 |
def generate_response_and_audio(audio_bytes: bytes, state: AppState):
|
| 55 |
if state.client is None:
|
|
@@ -58,7 +65,7 @@ def generate_response_and_audio(audio_bytes: bytes, state: AppState):
|
|
| 58 |
format_ = state.output_format
|
| 59 |
bitrate = 128 if format_ == "mp3" else 32 # Higher bitrate for MP3, lower for OPUS
|
| 60 |
audio_data = base64.b64encode(audio_bytes).decode()
|
| 61 |
-
|
| 62 |
try:
|
| 63 |
stream = state.client.chat.completions.create(
|
| 64 |
extra_body={
|
|
@@ -90,9 +97,6 @@ def generate_response_and_audio(audio_bytes: bytes, state: AppState):
|
|
| 90 |
|
| 91 |
final_audio = b''.join([base64.b64decode(a) for a in audios])
|
| 92 |
|
| 93 |
-
state.conversation.append({"role": "user", "content": "Audio input"})
|
| 94 |
-
state.conversation.append({"role": "assistant", "content": full_response})
|
| 95 |
-
|
| 96 |
yield full_response, final_audio, state
|
| 97 |
|
| 98 |
except Exception as e:
|
|
@@ -101,7 +105,7 @@ def generate_response_and_audio(audio_bytes: bytes, state: AppState):
|
|
| 101 |
def response(state: AppState):
|
| 102 |
if state.stream is None or len(state.stream) == 0:
|
| 103 |
return None, None, state
|
| 104 |
-
|
| 105 |
audio_buffer = io.BytesIO()
|
| 106 |
segment = AudioSegment(
|
| 107 |
state.stream.tobytes(),
|
|
@@ -112,7 +116,7 @@ def response(state: AppState):
|
|
| 112 |
segment.export(audio_buffer, format="wav")
|
| 113 |
|
| 114 |
generator = generate_response_and_audio(audio_buffer.getvalue(), state)
|
| 115 |
-
|
| 116 |
# Process the generator to get the final results
|
| 117 |
final_text = ""
|
| 118 |
final_audio = None
|
|
@@ -122,15 +126,23 @@ def response(state: AppState):
|
|
| 122 |
state = updated_state
|
| 123 |
|
| 124 |
# Update the chatbot with the final conversation
|
| 125 |
-
|
| 126 |
-
|
|
|
|
| 127 |
# Reset the audio stream for the next interaction
|
| 128 |
state.stream = None
|
| 129 |
-
state.
|
| 130 |
-
|
| 131 |
-
|
|
|
|
| 132 |
return chatbot_output, final_audio, state
|
| 133 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 134 |
def set_api_key(api_key, state):
|
| 135 |
if not api_key:
|
| 136 |
raise gr.Error("Please enter a valid API key.")
|
|
@@ -145,19 +157,19 @@ with gr.Blocks() as demo:
|
|
| 145 |
with gr.Row():
|
| 146 |
api_key_input = gr.Textbox(type="password", label="Enter your Lepton API Key")
|
| 147 |
set_key_button = gr.Button("Set API Key")
|
| 148 |
-
|
| 149 |
api_key_status = gr.Textbox(label="API Key Status", interactive=False)
|
| 150 |
-
|
| 151 |
with gr.Row():
|
| 152 |
format_dropdown = gr.Dropdown(choices=["mp3", "opus"], value="mp3", label="Output Audio Format")
|
| 153 |
-
|
| 154 |
with gr.Row():
|
| 155 |
with gr.Column():
|
| 156 |
input_audio = gr.Audio(label="Input Audio", sources="microphone", type="numpy")
|
| 157 |
with gr.Column():
|
| 158 |
chatbot = gr.Chatbot(label="Conversation", type="messages")
|
| 159 |
output_audio = gr.Audio(label="Output Audio", autoplay=True)
|
| 160 |
-
|
| 161 |
state = gr.State(AppState())
|
| 162 |
|
| 163 |
set_key_button.click(set_api_key, inputs=[api_key_input, state], outputs=[api_key_status, state])
|
|
@@ -170,11 +182,25 @@ with gr.Blocks() as demo:
|
|
| 170 |
stream_every=0.25, # Reduced to make it more responsive
|
| 171 |
time_limit=60, # Increased to allow for longer messages
|
| 172 |
)
|
| 173 |
-
|
| 174 |
respond = input_audio.stop_recording(
|
| 175 |
response,
|
| 176 |
[state],
|
| 177 |
[chatbot, output_audio, state]
|
| 178 |
)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 179 |
|
| 180 |
-
demo.launch()
|
|
|
|
| 3 |
import io
|
| 4 |
from pydub import AudioSegment
|
| 5 |
import tempfile
|
|
|
|
|
|
|
| 6 |
import openai
|
| 7 |
import time
|
| 8 |
from dataclasses import dataclass, field
|
|
|
|
| 12 |
class AppState:
|
| 13 |
stream: np.ndarray | None = None
|
| 14 |
sampling_rate: int = 0
|
| 15 |
+
pause_detected: bool = False
|
|
|
|
| 16 |
conversation: list = field(default_factory=list)
|
| 17 |
client: openai.OpenAI = None
|
| 18 |
output_format: str = "mp3"
|
| 19 |
+
stopped: bool = False
|
| 20 |
|
| 21 |
# Global lock for thread safety
|
| 22 |
state_lock = Lock()
|
|
|
|
| 27 |
api_key=api_key
|
| 28 |
)
|
| 29 |
|
| 30 |
+
def determine_pause(audio, sampling_rate, state):
|
| 31 |
+
# Take the last 1 second of audio
|
| 32 |
+
pause_length = int(sampling_rate * 1) # 1 second
|
| 33 |
+
if len(audio) < pause_length:
|
| 34 |
+
return False
|
| 35 |
+
last_audio = audio[-pause_length:]
|
| 36 |
+
amplitude = np.abs(last_audio)
|
| 37 |
+
|
| 38 |
+
# Calculate the average amplitude in the last 1 second
|
| 39 |
+
avg_amplitude = np.mean(amplitude)
|
| 40 |
+
silence_threshold = 0.01 # Adjust this threshold as needed
|
| 41 |
+
if avg_amplitude < silence_threshold:
|
| 42 |
+
return True
|
| 43 |
+
else:
|
| 44 |
+
return False
|
| 45 |
+
|
| 46 |
def process_audio(audio: tuple, state: AppState):
|
| 47 |
if state.stream is None:
|
| 48 |
state.stream = audio[1]
|
| 49 |
state.sampling_rate = audio[0]
|
|
|
|
| 50 |
else:
|
| 51 |
state.stream = np.concatenate((state.stream, audio[1]))
|
| 52 |
|
| 53 |
+
pause_detected = determine_pause(state.stream, state.sampling_rate, state)
|
| 54 |
+
state.pause_detected = pause_detected
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 55 |
|
| 56 |
+
if state.pause_detected:
|
|
|
|
| 57 |
return gr.Audio(recording=False), state
|
| 58 |
+
else:
|
| 59 |
+
return None, state
|
| 60 |
|
| 61 |
def generate_response_and_audio(audio_bytes: bytes, state: AppState):
|
| 62 |
if state.client is None:
|
|
|
|
| 65 |
format_ = state.output_format
|
| 66 |
bitrate = 128 if format_ == "mp3" else 32 # Higher bitrate for MP3, lower for OPUS
|
| 67 |
audio_data = base64.b64encode(audio_bytes).decode()
|
| 68 |
+
|
| 69 |
try:
|
| 70 |
stream = state.client.chat.completions.create(
|
| 71 |
extra_body={
|
|
|
|
| 97 |
|
| 98 |
final_audio = b''.join([base64.b64decode(a) for a in audios])
|
| 99 |
|
|
|
|
|
|
|
|
|
|
| 100 |
yield full_response, final_audio, state
|
| 101 |
|
| 102 |
except Exception as e:
|
|
|
|
| 105 |
def response(state: AppState):
|
| 106 |
if state.stream is None or len(state.stream) == 0:
|
| 107 |
return None, None, state
|
| 108 |
+
|
| 109 |
audio_buffer = io.BytesIO()
|
| 110 |
segment = AudioSegment(
|
| 111 |
state.stream.tobytes(),
|
|
|
|
| 116 |
segment.export(audio_buffer, format="wav")
|
| 117 |
|
| 118 |
generator = generate_response_and_audio(audio_buffer.getvalue(), state)
|
| 119 |
+
|
| 120 |
# Process the generator to get the final results
|
| 121 |
final_text = ""
|
| 122 |
final_audio = None
|
|
|
|
| 126 |
state = updated_state
|
| 127 |
|
| 128 |
# Update the chatbot with the final conversation
|
| 129 |
+
state.conversation.append({"role": "user", "content": "Audio input"})
|
| 130 |
+
state.conversation.append({"role": "assistant", "content": final_text})
|
| 131 |
+
|
| 132 |
# Reset the audio stream for the next interaction
|
| 133 |
state.stream = None
|
| 134 |
+
state.pause_detected = False
|
| 135 |
+
|
| 136 |
+
chatbot_output = state.conversation[-2:] # Get the last two messages
|
| 137 |
+
|
| 138 |
return chatbot_output, final_audio, state
|
| 139 |
|
| 140 |
+
def start_recording_user(state: AppState):
|
| 141 |
+
if not state.stopped:
|
| 142 |
+
return gr.Audio(recording=True)
|
| 143 |
+
else:
|
| 144 |
+
return gr.Audio(recording=False)
|
| 145 |
+
|
| 146 |
def set_api_key(api_key, state):
|
| 147 |
if not api_key:
|
| 148 |
raise gr.Error("Please enter a valid API key.")
|
|
|
|
| 157 |
with gr.Row():
|
| 158 |
api_key_input = gr.Textbox(type="password", label="Enter your Lepton API Key")
|
| 159 |
set_key_button = gr.Button("Set API Key")
|
| 160 |
+
|
| 161 |
api_key_status = gr.Textbox(label="API Key Status", interactive=False)
|
| 162 |
+
|
| 163 |
with gr.Row():
|
| 164 |
format_dropdown = gr.Dropdown(choices=["mp3", "opus"], value="mp3", label="Output Audio Format")
|
| 165 |
+
|
| 166 |
with gr.Row():
|
| 167 |
with gr.Column():
|
| 168 |
input_audio = gr.Audio(label="Input Audio", sources="microphone", type="numpy")
|
| 169 |
with gr.Column():
|
| 170 |
chatbot = gr.Chatbot(label="Conversation", type="messages")
|
| 171 |
output_audio = gr.Audio(label="Output Audio", autoplay=True)
|
| 172 |
+
|
| 173 |
state = gr.State(AppState())
|
| 174 |
|
| 175 |
set_key_button.click(set_api_key, inputs=[api_key_input, state], outputs=[api_key_status, state])
|
|
|
|
| 182 |
stream_every=0.25, # Reduced to make it more responsive
|
| 183 |
time_limit=60, # Increased to allow for longer messages
|
| 184 |
)
|
| 185 |
+
|
| 186 |
respond = input_audio.stop_recording(
|
| 187 |
response,
|
| 188 |
[state],
|
| 189 |
[chatbot, output_audio, state]
|
| 190 |
)
|
| 191 |
+
# Update the chatbot with the final conversation
|
| 192 |
+
respond.then(lambda s: s.conversation, [state], [chatbot])
|
| 193 |
+
|
| 194 |
+
# Automatically restart recording after the assistant's response
|
| 195 |
+
restart = output_audio.stop(
|
| 196 |
+
start_recording_user,
|
| 197 |
+
[state],
|
| 198 |
+
[input_audio]
|
| 199 |
+
)
|
| 200 |
+
|
| 201 |
+
# Add a "Stop Conversation" button
|
| 202 |
+
cancel = gr.Button("Stop Conversation", variant="stop")
|
| 203 |
+
cancel.click(lambda: (AppState(stopped=True), gr.Audio(recording=False)), None,
|
| 204 |
+
[state, input_audio], cancels=[respond, restart])
|
| 205 |
|
| 206 |
+
demo.launch()
|