Spaces:
Runtime error
Runtime error
better formatting
Browse files- app.py +11 -22
- charles_actor.py +26 -23
app.py
CHANGED
|
@@ -39,34 +39,24 @@ if not ray.is_initialized():
|
|
| 39 |
|
| 40 |
|
| 41 |
async def main():
|
| 42 |
-
|
| 43 |
-
system_one_audio_status = st.empty()
|
| 44 |
-
|
| 45 |
-
|
| 46 |
-
system_one_audio_status.write("Initializing streaming")
|
| 47 |
-
|
| 48 |
-
# system_one_audio_output = st.empty()
|
| 49 |
-
# system_one_video_output = st.empty()
|
| 50 |
-
# system_one_audio_history = []
|
| 51 |
|
| 52 |
col1, col2 = st.columns(2)
|
| 53 |
|
| 54 |
with col1:
|
| 55 |
-
|
| 56 |
-
|
|
|
|
|
|
|
|
|
|
| 57 |
charles_actor_debug_output = st.empty()
|
| 58 |
environment_state_ouput = st.empty()
|
| 59 |
|
| 60 |
-
# Initialize resources if not already done
|
| 61 |
-
system_one_audio_status.write("Initializing streaming")
|
| 62 |
-
if "streamlit_av_queue" not in st.session_state:
|
| 63 |
-
from streamlit_av_queue import StreamlitAVQueue
|
| 64 |
-
st.session_state.streamlit_av_queue = StreamlitAVQueue()
|
| 65 |
-
|
| 66 |
-
system_one_audio_status.write("resources referecned")
|
| 67 |
-
|
| 68 |
-
system_one_audio_status.write("Initializing webrtc_streamer")
|
| 69 |
with col2:
|
|
|
|
|
|
|
|
|
|
|
|
|
| 70 |
playing = st.checkbox("Playing", value=True)
|
| 71 |
webrtc_ctx = webrtc_streamer(
|
| 72 |
key="charles",
|
|
@@ -87,12 +77,11 @@ async def main():
|
|
| 87 |
rtc_configuration={"iceServers": get_ice_servers()},
|
| 88 |
async_processing=True,
|
| 89 |
)
|
|
|
|
| 90 |
|
| 91 |
if not webrtc_ctx.state.playing:
|
| 92 |
exit
|
| 93 |
|
| 94 |
-
system_one_audio_status.write("Initializing speech")
|
| 95 |
-
|
| 96 |
from charles_actor import CharlesActor
|
| 97 |
charles_actor = None
|
| 98 |
|
|
|
|
| 39 |
|
| 40 |
|
| 41 |
async def main():
|
| 42 |
+
st.title("Project Charles")
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 43 |
|
| 44 |
col1, col2 = st.columns(2)
|
| 45 |
|
| 46 |
with col1:
|
| 47 |
+
nested_col1, nested_col2 = st.columns(2)
|
| 48 |
+
with nested_col1:
|
| 49 |
+
listening = st.checkbox("Listen", value=True)
|
| 50 |
+
with nested_col2:
|
| 51 |
+
looking = st.checkbox("Look", value=False)
|
| 52 |
charles_actor_debug_output = st.empty()
|
| 53 |
environment_state_ouput = st.empty()
|
| 54 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 55 |
with col2:
|
| 56 |
+
if "streamlit_av_queue" not in st.session_state:
|
| 57 |
+
from streamlit_av_queue import StreamlitAVQueue
|
| 58 |
+
st.session_state.streamlit_av_queue = StreamlitAVQueue()
|
| 59 |
+
|
| 60 |
playing = st.checkbox("Playing", value=True)
|
| 61 |
webrtc_ctx = webrtc_streamer(
|
| 62 |
key="charles",
|
|
|
|
| 77 |
rtc_configuration={"iceServers": get_ice_servers()},
|
| 78 |
async_processing=True,
|
| 79 |
)
|
| 80 |
+
system_one_audio_status = st.markdown("Initializing streaming")
|
| 81 |
|
| 82 |
if not webrtc_ctx.state.playing:
|
| 83 |
exit
|
| 84 |
|
|
|
|
|
|
|
| 85 |
from charles_actor import CharlesActor
|
| 86 |
charles_actor = None
|
| 87 |
|
charles_actor.py
CHANGED
|
@@ -61,13 +61,19 @@ class CharlesActor:
|
|
| 61 |
await self._initalize_resources()
|
| 62 |
|
| 63 |
debug_output_history = []
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 64 |
def add_debug_output(output):
|
| 65 |
debug_output_history.append(output)
|
| 66 |
if len(debug_output_history) > 10:
|
| 67 |
debug_output_history.pop(0)
|
| 68 |
-
|
| 69 |
-
table_content += "\n".join([f"| {item} |" for item in reversed(debug_output_history)])
|
| 70 |
-
self._charles_actor_debug_output = table_content
|
| 71 |
|
| 72 |
self._state = "Waiting for input"
|
| 73 |
total_video_frames = 0
|
|
@@ -123,12 +129,12 @@ class CharlesActor:
|
|
| 123 |
|
| 124 |
if speaker_finished and len(prompt) > 0 and prompt not in prompts_to_ignore:
|
| 125 |
print(f"Prompt: {prompt}")
|
| 126 |
-
|
| 127 |
for i, response in enumerate(current_responses):
|
| 128 |
-
line
|
| 129 |
-
line += f"{response} [{speech_chunks_per_response[i]}]"
|
| 130 |
-
|
| 131 |
-
|
| 132 |
add_debug_output(line)
|
| 133 |
add_debug_output(f"π¨ {prompt}")
|
| 134 |
current_responses = []
|
|
@@ -156,23 +162,20 @@ class CharlesActor:
|
|
| 156 |
response_id = chunk['llm_sentence_id']
|
| 157 |
speech_chunks_per_response[response_id] += 1
|
| 158 |
|
| 159 |
-
|
| 160 |
-
|
| 161 |
-
if len(robot_preview_text) > 0:
|
| 162 |
-
debug_output_history_copy.append(robot_preview_text)
|
| 163 |
-
lines = []
|
| 164 |
for i, response in enumerate(current_responses):
|
| 165 |
-
line
|
| 166 |
-
line += f"
|
| 167 |
-
|
| 168 |
-
|
| 169 |
-
|
|
|
|
| 170 |
if len(human_preview_text) > 0:
|
| 171 |
-
|
| 172 |
-
if len(
|
| 173 |
-
|
| 174 |
-
|
| 175 |
-
self._charles_actor_debug_output = table_content
|
| 176 |
|
| 177 |
|
| 178 |
await asyncio.sleep(0.01)
|
|
|
|
| 61 |
await self._initalize_resources()
|
| 62 |
|
| 63 |
debug_output_history = []
|
| 64 |
+
|
| 65 |
+
def render_debug_output(list_of_strings):
|
| 66 |
+
table_content = "##### Chat history\n"
|
| 67 |
+
for item in reversed(list_of_strings):
|
| 68 |
+
# table_content += f"\n```markdown\n{item}\n```\n"
|
| 69 |
+
table_content += f"\n{item}\n"
|
| 70 |
+
self._charles_actor_debug_output = table_content
|
| 71 |
+
|
| 72 |
def add_debug_output(output):
|
| 73 |
debug_output_history.append(output)
|
| 74 |
if len(debug_output_history) > 10:
|
| 75 |
debug_output_history.pop(0)
|
| 76 |
+
render_debug_output(debug_output_history)
|
|
|
|
|
|
|
| 77 |
|
| 78 |
self._state = "Waiting for input"
|
| 79 |
total_video_frames = 0
|
|
|
|
| 129 |
|
| 130 |
if speaker_finished and len(prompt) > 0 and prompt not in prompts_to_ignore:
|
| 131 |
print(f"Prompt: {prompt}")
|
| 132 |
+
line = ""
|
| 133 |
for i, response in enumerate(current_responses):
|
| 134 |
+
line += "π€ " if len(line) == 0 else ""
|
| 135 |
+
# line += f"{response} [{speech_chunks_per_response[i]}] \n"
|
| 136 |
+
line += f"[{speech_chunks_per_response[i]}] {response} \n"
|
| 137 |
+
if len(line) > 0:
|
| 138 |
add_debug_output(line)
|
| 139 |
add_debug_output(f"π¨ {prompt}")
|
| 140 |
current_responses = []
|
|
|
|
| 162 |
response_id = chunk['llm_sentence_id']
|
| 163 |
speech_chunks_per_response[response_id] += 1
|
| 164 |
|
| 165 |
+
list_of_strings = debug_output_history.copy()
|
| 166 |
+
line = ""
|
|
|
|
|
|
|
|
|
|
| 167 |
for i, response in enumerate(current_responses):
|
| 168 |
+
line += "π€ " if len(line) == 0 else ""
|
| 169 |
+
line += f"[{speech_chunks_per_response[i]}] {response} \n"
|
| 170 |
+
# line += f"{response} [{speech_chunks_per_response[i]}] \n"
|
| 171 |
+
if len(robot_preview_text) > 0:
|
| 172 |
+
line += robot_preview_text+" \n"
|
| 173 |
+
list_of_strings.append(line)
|
| 174 |
if len(human_preview_text) > 0:
|
| 175 |
+
list_of_strings.append(human_preview_text)
|
| 176 |
+
if len(list_of_strings) > 10:
|
| 177 |
+
list_of_strings.pop(0)
|
| 178 |
+
render_debug_output(list_of_strings)
|
|
|
|
| 179 |
|
| 180 |
|
| 181 |
await asyncio.sleep(0.01)
|