klas/t6_mem0
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Formatting (#2750)

Browse Source
This commit is contained in:
Dev Khant
2025-05-22 01:17:29 +05:30
committed by GitHub
parent dff91154a7
commit d85fcda037
71 changed files with 1391 additions and 1823 deletions
Download Patch File Download Diff File Expand all files Collapse all files

10
examples/graph-db-demo/memgraph-example.ipynb
View File

@@ -56,9 +56,7 @@
"\n",
"import os\n",
"\n",
"os.environ[\"OPENAI_API_KEY\"] = (\n",
" \"\"\n",
")"
"os.environ[\"OPENAI_API_KEY\"] = \"\""
]
},
{
@@ -149,7 +147,7 @@
" \"role\": \"assistant\",\n",
" \"content\": \"Got it! I'll avoid thriller recommendations and suggest sci-fi movies in the future.\",\n",
" },\n",
"]\n"
"]"
]
},
{
@@ -166,9 +164,7 @@
"outputs": [],
"source": [
"# Store inferred memories (default behavior)\n",
"result = m.add(\n",
" messages, user_id=\"alice\", metadata={\"category\": \"movie_recommendations\"}\n",
")"
"result = m.add(messages, user_id=\"alice\", metadata={\"category\": \"movie_recommendations\"})"
]
},
{

107
examples/misc/fitness_checker.py
View File

@@ -20,19 +20,19 @@ agent = Agent(
name="Fitness Agent",
model=OpenAIChat(id="gpt-4o"),
description="You are a helpful fitness assistant who remembers past logs and gives personalized suggestions for Anish's training and diet.",
markdown=True
markdown=True,
)
# Store user preferences as memory
def store_user_preferences(conversation: list, user_id: str = USER_ID):
"""Store user preferences from conversation history"""
memory_client.add(conversation, user_id=user_id, output_format='v1.1')
memory_client.add(conversation, user_id=user_id, output_format="v1.1")
# Memory-aware assistant function
def fitness_coach(user_input: str, user_id: str = USER_ID):
memories = memory_client.search(user_input, user_id=user_id) # Search relevant memories bases on user query
memories = memory_client.search(user_input, user_id=user_id) # Search relevant memories bases on user query
memory_context = "\n".join(f"- {m['memory']}" for m in memories)
prompt = f"""You are a fitness assistant who helps Anish with his training, recovery, and diet. You have long-term memory of his health, routines, preferences, and past conversations.
@@ -48,113 +48,66 @@ User query:
memory_client.add(f"User: {user_input}\nAssistant: {response.content}", user_id=user_id)
return response.content
# --------------------------------------------------
# Store user preferences and memories
messages = [
{
"role": "user",
"content": "Hi, I’m Anish. I'm 26 years old, 5'10\", and weigh 72kg. I started working out 6 months ago with the goal of building lean muscle."
"content": "Hi, I’m Anish. I'm 26 years old, 5'10\", and weigh 72kg. I started working out 6 months ago with the goal of building lean muscle.",
},
{
"role": "assistant",
"content": "Got it — you're 26, 5'10\", 72kg, and on a lean muscle journey. Started gym 6 months ago."
"content": "Got it — you're 26, 5'10\", 72kg, and on a lean muscle journey. Started gym 6 months ago.",
},
{
"role": "user",
"content": "I follow a push-pull-legs routine and train 5 times a week. My rest days are Wednesday and Sunday."
"content": "I follow a push-pull-legs routine and train 5 times a week. My rest days are Wednesday and Sunday.",
},
{
"role": "assistant",
"content": "Understood — push-pull-legs split, training 5x/week with rest on Wednesdays and Sundays."
"content": "Understood — push-pull-legs split, training 5x/week with rest on Wednesdays and Sundays.",
},
{"role": "user", "content": "After push days, I usually eat high-protein and moderate-carb meals to recover."},
{"role": "assistant", "content": "Noted — high-protein, moderate-carb meals after push workouts."},
{"role": "user", "content": "For pull days, I take whey protein and eat a banana after training."},
{"role": "assistant", "content": "Logged — whey protein and banana post pull workouts."},
{"role": "user", "content": "On leg days, I make sure to have complex carbs like rice or oats."},
{"role": "assistant", "content": "Noted — complex carbs like rice and oats are part of your leg day meals."},
{
"role": "user",
"content": "After push days, I usually eat high-protein and moderate-carb meals to recover."
},
{
"role": "assistant",
"content": "Noted — high-protein, moderate-carb meals after push workouts."
"content": "I often feel sore after leg days, so I use turmeric milk and magnesium to help with recovery.",
},
{"role": "assistant", "content": "I'll remember turmeric milk and magnesium as part of your leg day recovery."},
{
"role": "user",
"content": "For pull days, I take whey protein and eat a banana after training."
"content": "Last push day, I did 3x8 bench press at 60kg, 4x12 overhead press, and dips. Felt fatigued after.",
},
{
"role": "assistant",
"content": "Logged — whey protein and banana post pull workouts."
"content": "Push day logged — 60kg bench, overhead press, dips. You felt fatigued afterward.",
},
{"role": "user", "content": "I prefer light dinners post-workout like tofu, soup, and vegetables."},
{"role": "assistant", "content": "Got it — light dinners post-workout: tofu, soup, and veggies."},
{
"role": "user",
"content": "On leg days, I make sure to have complex carbs like rice or oats."
},
{
"role": "assistant",
"content": "Noted — complex carbs like rice and oats are part of your leg day meals."
"content": "I have mild lactose intolerance, so I avoid dairy. I use almond milk or lactose-free whey.",
},
{"role": "assistant", "content": "Understood — avoiding regular dairy, using almond milk and lactose-free whey."},
{
"role": "user",
"content": "I often feel sore after leg days, so I use turmeric milk and magnesium to help with recovery."
"content": "I get occasional knee pain, so I avoid deep squats and do more hamstring curls and glute bridges on leg days.",
},
{
"role": "assistant",
"content": "I'll remember turmeric milk and magnesium as part of your leg day recovery."
},
{
"role": "user",
"content": "Last push day, I did 3x8 bench press at 60kg, 4x12 overhead press, and dips. Felt fatigued after."
},
{
"role": "assistant",
"content": "Push day logged — 60kg bench, overhead press, dips. You felt fatigued afterward."
},
{
"role": "user",
"content": "I prefer light dinners post-workout like tofu, soup, and vegetables."
},
{
"role": "assistant",
"content": "Got it — light dinners post-workout: tofu, soup, and veggies."
},
{
"role": "user",
"content": "I have mild lactose intolerance, so I avoid dairy. I use almond milk or lactose-free whey."
},
{
"role": "assistant",
"content": "Understood — avoiding regular dairy, using almond milk and lactose-free whey."
},
{
"role": "user",
"content": "I get occasional knee pain, so I avoid deep squats and do more hamstring curls and glute bridges on leg days."
},
{
"role": "assistant",
"content": "Noted — due to knee discomfort, you substitute deep squats with curls and glute bridges."
},
{
"role": "user",
"content": "I track sleep and notice poor performance when I sleep less than 6 hours."
},
{
"role": "assistant",
"content": "Logged — performance drops when you get under 6 hours of sleep."
},
{
"role": "user",
"content": "I take magnesium supplements to help with muscle recovery and sleep quality."
},
{
"role": "assistant",
"content": "Remembered — magnesium helps you with recovery and sleep."
},
{
"role": "user",
"content": "I avoid caffeine after 4 PM because it affects my sleep."
},
{
"role": "assistant",
"content": "Got it — you avoid caffeine post-4 PM to protect your sleep."
"content": "Noted — due to knee discomfort, you substitute deep squats with curls and glute bridges.",
},
{"role": "user", "content": "I track sleep and notice poor performance when I sleep less than 6 hours."},
{"role": "assistant", "content": "Logged — performance drops when you get under 6 hours of sleep."},
{"role": "user", "content": "I take magnesium supplements to help with muscle recovery and sleep quality."},
{"role": "assistant", "content": "Remembered — magnesium helps you with recovery and sleep."},
{"role": "user", "content": "I avoid caffeine after 4 PM because it affects my sleep."},
{"role": "assistant", "content": "Got it — you avoid caffeine post-4 PM to protect your sleep."},
]
store_user_preferences(messages)

77
examples/misc/healthcare_assistant_google_adk.py
View File

@@ -1,9 +1,11 @@
import asyncio
import warnings
from google.adk.agents import Agent
from google.adk.sessions import InMemorySessionService
from google.adk.runners import Runner
from google.adk.sessions import InMemorySessionService
from google.genai import types
from mem0 import MemoryClient
warnings.filterwarnings("ignore", category=DeprecationWarning)
@@ -19,14 +21,14 @@ def save_patient_info(information: str) -> dict:
print(f"Storing patient information: {information[:30]}...")
# Get user_id from session state or use default
user_id = getattr(save_patient_info, 'user_id', 'default_user')
user_id = getattr(save_patient_info, "user_id", "default_user")
# Store in Mem0
response = mem0_client.add(
mem0_client.add(
[{"role": "user", "content": information}],
user_id=user_id,
run_id="healthcare_session",
metadata={"type": "patient_information"}
metadata={"type": "patient_information"},
)
return {"status": "success", "message": "Information saved"}
@@ -37,7 +39,7 @@ def retrieve_patient_info(query: str) -> str:
print(f"Searching for patient information: {query}")
# Get user_id from session state or use default
user_id = getattr(retrieve_patient_info, 'user_id', 'default_user')
user_id = getattr(retrieve_patient_info, "user_id", "default_user")
# Search Mem0
results = mem0_client.search(
@@ -45,7 +47,7 @@ def retrieve_patient_info(query: str) -> str:
user_id=user_id,
run_id="healthcare_session",
limit=5,
threshold=0.7 # Higher threshold for more relevant results
threshold=0.7, # Higher threshold for more relevant results
)
if not results:
@@ -65,7 +67,7 @@ def schedule_appointment(date: str, time: str, reason: str) -> dict:
"status": "success",
"appointment_id": appointment_id,
"confirmation": f"Appointment scheduled for {date} at {time} for {reason}",
"message": "Please arrive 15 minutes early to complete paperwork."
"message": "Please arrive 15 minutes early to complete paperwork.",
}
@@ -89,7 +91,7 @@ IMPORTANT GUIDELINES:
- For serious symptoms, always recommend consulting a healthcare professional.
- Keep all patient information confidential.
""",
tools=[save_patient_info, retrieve_patient_info, schedule_appointment]
tools=[save_patient_info, retrieve_patient_info, schedule_appointment],
)
# Set Up Session and Runner
@@ -101,18 +103,10 @@ USER_ID = "Alex"
SESSION_ID = "session_001"
# Create a session
session = session_service.create_session(
app_name=APP_NAME,
user_id=USER_ID,
session_id=SESSION_ID
)
session = session_service.create_session(app_name=APP_NAME, user_id=USER_ID, session_id=SESSION_ID)
# Create the runner
runner = Runner(
agent=healthcare_agent,
app_name=APP_NAME,
session_service=session_service
)
runner = Runner(agent=healthcare_agent, app_name=APP_NAME, session_service=session_service)
# Interact with the Healthcare Assistant
@@ -121,21 +115,14 @@ async def call_agent_async(query, runner, user_id, session_id):
print(f"\n>>> Patient: {query}")
# Format the user's message
content = types.Content(
role='user',
parts=[types.Part(text=query)]
)
content = types.Content(role="user", parts=[types.Part(text=query)])
# Set user_id for tools to access
save_patient_info.user_id = user_id
retrieve_patient_info.user_id = user_id
# Run the agent
async for event in runner.run_async(
user_id=user_id,
session_id=session_id,
new_message=content
):
async for event in runner.run_async(user_id=user_id, session_id=session_id, new_message=content):
if event.is_final_response():
if event.content and event.content.parts:
response = event.content.parts[0].text
@@ -152,7 +139,7 @@ async def run_conversation():
"Hi, I'm Alex. I've been having headaches for the past week, and I have a penicillin allergy.",
runner=runner,
user_id=USER_ID,
session_id=SESSION_ID
session_id=SESSION_ID,
)
# Request for health information
@@ -160,7 +147,7 @@ async def run_conversation():
"Can you tell me more about what might be causing my headaches?",
runner=runner,
user_id=USER_ID,
session_id=SESSION_ID
session_id=SESSION_ID,
)
# Schedule an appointment
@@ -168,15 +155,12 @@ async def run_conversation():
"I think I should see a doctor. Can you help me schedule an appointment for next Monday at 2pm?",
runner=runner,
user_id=USER_ID,
session_id=SESSION_ID
session_id=SESSION_ID,
)
# Test memory - should remember patient name, symptoms, and allergy
await call_agent_async(
"What medications should I avoid for my headaches?",
runner=runner,
user_id=USER_ID,
session_id=SESSION_ID
"What medications should I avoid for my headaches?", runner=runner, user_id=USER_ID, session_id=SESSION_ID
)
@@ -191,37 +175,28 @@ async def interactive_mode():
session_id = f"session_{hash(patient_id) % 1000:03d}"
# Create session for this user
session = session_service.create_session(
app_name=APP_NAME,
user_id=patient_id,
session_id=session_id
)
session_service.create_session(app_name=APP_NAME, user_id=patient_id, session_id=session_id)
print(f"\nStarting conversation with patient ID: {patient_id}")
print("Type your message and press Enter.")
while True:
user_input = input("\n>>> Patient: ").strip()
if user_input.lower() in ['exit', 'quit', 'bye']:
if user_input.lower() in ["exit", "quit", "bye"]:
print("Ending conversation. Thank you!")
break
await call_agent_async(
user_input,
runner=runner,
user_id=patient_id,
session_id=session_id
)
await call_agent_async(user_input, runner=runner, user_id=patient_id, session_id=session_id)
# Main execution
if __name__ == "__main__":
import argparse
parser = argparse.ArgumentParser(description='Healthcare Assistant with Memory')
parser.add_argument('--demo', action='store_true', help='Run the demo conversation')
parser.add_argument('--interactive', action='store_true', help='Run in interactive mode')
parser.add_argument('--patient-id', type=str, default=USER_ID, help='Patient ID for the conversation')
parser = argparse.ArgumentParser(description="Healthcare Assistant with Memory")
parser.add_argument("--demo", action="store_true", help="Run the demo conversation")
parser.add_argument("--interactive", action="store_true", help="Run in interactive mode")
parser.add_argument("--patient-id", type=str, default=USER_ID, help="Patient ID for the conversation")
args = parser.parse_args()
if args.demo:
@@ -231,5 +206,3 @@ if __name__ == "__main__":
else:
# Default to demo mode if no arguments provided
asyncio.run(run_conversation())

30
examples/misc/movie_recommendation_grok3.py
View File

@@ -16,26 +16,21 @@ from mem0 import Memory
# Configure Mem0 with Grok 3 and Qdrant
config = {
"vector_store": {
"provider": "qdrant",
"config": {
"embedding_model_dims": 384
}
},
"vector_store": {"provider": "qdrant", "config": {"embedding_model_dims": 384}},
"llm": {
"provider": "xai",
"config": {
"model": "grok-3-beta",
"temperature": 0.1,
"max_tokens": 2000,
}
},
},
"embedder": {
"provider": "huggingface",
"config": {
"model": "all-MiniLM-L6-v2" # open embedding model
}
}
},
},
}
# Instantiate memory layer
@@ -57,20 +52,14 @@ def recommend_movie_with_memory(user_id: str, user_query: str):
prompt += f"\nPreviously, the user mentioned: {past_memories}"
# Generate movie recommendation using Grok 3
response = grok_client.chat.completions.create(
model="grok-3-beta",
messages=[
{"role": "user", "content": prompt}
]
)
response = grok_client.chat.completions.create(model="grok-3-beta", messages=[{"role": "user", "content": prompt}])
recommendation = response.choices[0].message.content
# Store conversation in memory
memory.add(
[{"role": "user", "content": user_query},
{"role": "assistant", "content": recommendation}],
[{"role": "user", "content": user_query}, {"role": "assistant", "content": recommendation}],
user_id=user_id,
metadata={"category": "movie"}
metadata={"category": "movie"},
)
return recommendation
@@ -81,10 +70,11 @@ if __name__ == "__main__":
user_id = "arshi"
recommend_movie_with_memory(user_id, "I'm looking for a movie to watch tonight. Any suggestions?")
# OUTPUT: You have watched Intersteller last weekend and you don't like horror movies, maybe you can watch "Purple Hearts" today.
recommend_movie_with_memory(user_id, "Can we skip the tearjerkers? I really enjoyed Notting Hill and Crazy Rich Asians.")
recommend_movie_with_memory(
user_id, "Can we skip the tearjerkers? I really enjoyed Notting Hill and Crazy Rich Asians."
)
# OUTPUT: Got it — no sad endings! You might enjoy "The Proposal" or "Love, Rosie". They’re both light-hearted romcoms with happy vibes.
recommend_movie_with_memory(user_id, "Any light-hearted movie I can watch after work today?")
# OUTPUT: Since you liked Crazy Rich Asians and The Proposal, how about "The Intern" or "Isn’t It Romantic"? Both are upbeat, funny, and perfect for relaxing.
recommend_movie_with_memory(user_id, "I’ve already watched The Intern. Something new maybe?")
# OUTPUT: No problem! Try "Your Place or Mine" - romcoms that match your taste and are tear-free!

31
examples/misc/personal_assistant_agno.py
View File

@@ -23,8 +23,8 @@ agent = Agent(
name="Personal Agent",
model=OpenAIChat(id="gpt-4o"),
description="You are a helpful personal agent that helps me with day to day activities."
"You can process both text and images.",
markdown=True
"You can process both text and images.",
markdown=True,
)
@@ -35,24 +35,16 @@ def chat_user(user_input: str = None, user_id: str = "user_123", image_path: str
base64_image = base64.b64encode(image_file.read()).decode("utf-8")
# First: the text message
text_msg = {
"role": "user",
"content": user_input
}
text_msg = {"role": "user", "content": user_input}
# Second: the image message
image_msg = {
"role": "user",
"content": {
"type": "image_url",
"image_url": {
"url": f"data:image/jpeg;base64,{base64_image}"
}
}
"content": {"type": "image_url", "image_url": {"url": f"data:image/jpeg;base64,{base64_image}"}},
}
# Send both as separate message objects
client.add([text_msg, image_msg], user_id=user_id, output_format='v1.1')
client.add([text_msg, image_msg], user_id=user_id, output_format="v1.1")
print("✅ Image uploaded and stored in memory.")
if user_input:
@@ -92,10 +84,13 @@ print(chat_user("When is my test?", user_id=user_id))
# OUTPUT: Your pilot's test is on your birthday, which is in five days. You're turning 25!
# Good luck with your preparations, and remember to take some time to relax amidst the studying.
print(chat_user("This is the picture of what I brought with me in the trip to Bahamas",
image_path="travel_items.jpeg", # this will be added to Mem0 memory
user_id=user_id))
print(chat_user("hey can you quickly tell me if brought my sunglasses to my trip, not able to find",
user_id=user_id))
print(
chat_user(
"This is the picture of what I brought with me in the trip to Bahamas",
image_path="travel_items.jpeg", # this will be added to Mem0 memory
user_id=user_id,
)
)
print(chat_user("hey can you quickly tell me if brought my sunglasses to my trip, not able to find", user_id=user_id))
# OUTPUT: Yes, you did bring your sunglasses on your trip to the Bahamas along with your laptop, face masks and other items..
# Since you can't find them now, perhaps check the pockets of jackets you wore or in your luggage compartments.

28
examples/misc/study_buddy.py
View File

@@ -7,6 +7,7 @@ In order to run this file, you need to set up your Mem0 API at Mem0 platform and
export OPENAI_API_KEY="your_openai_api_key"
export MEM0_API_KEY="your_mem0_api_key"
"""
import asyncio
from agents import Agent, Runner
@@ -23,25 +24,19 @@ study_agent = Agent(
- Identify topics the user has struggled with (e.g., "I'm confused", "this is hard")
- Help with spaced repetition by suggesting topics to revisit based on last review time
- Personalize answers using stored memories
- Summarize PDFs or notes the user uploads""")
- Summarize PDFs or notes the user uploads""",
)
# Upload and store PDF to Mem0
def upload_pdf(pdf_url: str, user_id: str):
pdf_message = {
"role": "user",
"content": {
"type": "pdf_url",
"pdf_url": {"url": pdf_url}
}
}
pdf_message = {"role": "user", "content": {"type": "pdf_url", "pdf_url": {"url": pdf_url}}}
client.add([pdf_message], user_id=user_id)
print("✅ PDF uploaded and processed into memory.")
# Main interaction loop with your personal study buddy
async def study_buddy(user_id: str, topic: str, user_input: str):
memories = client.search(f"{topic}", user_id=user_id)
memory_context = "n".join(f"- {m['memory']}" for m in memories)
@@ -56,9 +51,11 @@ Now respond to the user's new question or comment:
result = await Runner.run(study_agent, prompt)
response = result.final_output
client.add([
{"role": "user", "content": f'''Topic: {topic}nUser: {user_input}nnStudy Assistant: {response}'''}
], user_id=user_id, metadata={"topic": topic})
client.add(
[{"role": "user", "content": f"""Topic: {topic}nUser: {user_input}nnStudy Assistant: {response}"""}],
user_id=user_id,
metadata={"topic": topic},
)
return response
@@ -78,7 +75,12 @@ async def main():
# Demonstrate spaced repetition prompting
topic = "Momentum Conservation"
print(await study_buddy(user_id, topic, "I think we covered this last week. Is it time to review momentum conservation again?"))
print(
await study_buddy(
user_id, topic, "I think we covered this last week. Is it time to review momentum conservation again?"
)
)
if __name__ == "__main__":
asyncio.run(main())

47
examples/misc/voice_assistant_elevenlabs.py
View File

@@ -57,7 +57,7 @@ def initialize_memory():
},
{
"role": "user",
"content": "I prefer brief and concise responses without unnecessary explanations. I get frustrated when assistants are too wordy or repeat information I already know."
"content": "I prefer brief and concise responses without unnecessary explanations. I get frustrated when assistants are too wordy or repeat information I already know.",
},
{
"role": "assistant",
@@ -65,7 +65,7 @@ def initialize_memory():
},
{
"role": "user",
"content": "I like to listen to jazz music when I'm working, especially artists like Miles Davis and John Coltrane. I find it helps me focus and be more productive."
"content": "I like to listen to jazz music when I'm working, especially artists like Miles Davis and John Coltrane. I find it helps me focus and be more productive.",
},
{
"role": "assistant",
@@ -73,7 +73,7 @@ def initialize_memory():
},
{
"role": "user",
"content": "I usually wake up at 7 AM and prefer reminders for meetings 30 minutes in advance. My most productive hours are between 9 AM and noon, so I try to schedule important tasks during that time."
"content": "I usually wake up at 7 AM and prefer reminders for meetings 30 minutes in advance. My most productive hours are between 9 AM and noon, so I try to schedule important tasks during that time.",
},
{
"role": "assistant",
@@ -81,7 +81,7 @@ def initialize_memory():
},
{
"role": "user",
"content": "My favorite color is navy blue, and I prefer dark mode in all my apps. I'm allergic to peanuts, so please remind me to check ingredients when I ask about recipes or restaurants."
"content": "My favorite color is navy blue, and I prefer dark mode in all my apps. I'm allergic to peanuts, so please remind me to check ingredients when I ask about recipes or restaurants.",
},
{
"role": "assistant",
@@ -89,7 +89,7 @@ def initialize_memory():
},
{
"role": "user",
"content": "My partner's name is Jamie, and we have a golden retriever named Max who is 3 years old. My parents live in Chicago, and I try to visit them once every two months."
"content": "My partner's name is Jamie, and we have a golden retriever named Max who is 3 years old. My parents live in Chicago, and I try to visit them once every two months.",
},
{
"role": "assistant",
@@ -135,11 +135,11 @@ def record_audio(filename="input.wav", record_seconds=5):
stream.close()
p.terminate()
with wave.open(filename, 'wb') as wf:
with wave.open(filename, "wb") as wf:
wf.setnchannels(channels)
wf.setsampwidth(p.get_sample_size(fmt))
wf.setframerate(rate)
wf.writeframes(b''.join(frames))
wf.writeframes(b"".join(frames))
# ------------------ STT USING WHISPER ------------------
@@ -147,10 +147,7 @@ def transcribe_whisper(audio_path):
print("🔎 Transcribing with Whisper...")
try:
with open(audio_path, "rb") as audio_file:
transcript = openai_client.audio.transcriptions.create(
model="whisper-1",
file=audio_file
)
transcript = openai_client.audio.transcriptions.create(model="whisper-1", file=audio_file)
print(f"🗣️ You said: {transcript.text}")
return transcript.text
except Exception as e:
@@ -165,9 +162,7 @@ def get_agent_response(user_input):
try:
task = Task(
description=f"Respond to: {user_input}",
expected_output="A short and relevant reply.",
agent=voice_agent
description=f"Respond to: {user_input}", expected_output="A short and relevant reply.", agent=voice_agent
)
crew = Crew(
agents=[voice_agent],
@@ -175,22 +170,19 @@ def get_agent_response(user_input):
process=Process.sequential,
verbose=True,
memory=True,
memory_config={
"provider": "mem0",
"config": {"user_id": USER_ID}
}
memory_config={"provider": "mem0", "config": {"user_id": USER_ID}},
)
result = crew.kickoff()
# Extract the text response from the complex result object
if hasattr(result, 'raw'):
if hasattr(result, "raw"):
return result.raw
elif isinstance(result, dict) and 'raw' in result:
return result['raw']
elif isinstance(result, dict) and 'tasks_output' in result:
outputs = result['tasks_output']
elif isinstance(result, dict) and "raw" in result:
return result["raw"]
elif isinstance(result, dict) and "tasks_output" in result:
outputs = result["tasks_output"]
if outputs and isinstance(outputs, list) and len(outputs) > 0:
return outputs[0].get('raw', str(result))
return outputs[0].get("raw", str(result))
# Fallback to string representation if we can't extract the raw response
return str(result)
@@ -204,10 +196,7 @@ def get_agent_response(user_input):
def speak_response(text):
print(f"🤖 Agent: {text}")
audio = tts_client.text_to_speech.convert(
text=text,
voice_id="JBFqnCBsd6RMkjVDRZzb",
model_id="eleven_multilingual_v2",
output_format="mp3_44100_128"
text=text, voice_id="JBFqnCBsd6RMkjVDRZzb", model_id="eleven_multilingual_v2", output_format="mp3_44100_128"
)
play(audio)
@@ -220,7 +209,7 @@ def run_voice_agent():
record_audio(tmp_audio.name)
try:
user_text = transcribe_whisper(tmp_audio.name)
if user_text.lower() in ['exit', 'quit', 'stop']:
if user_text.lower() in ["exit", "quit", "stop"]:
print("👋 Exiting.")
break
response = get_agent_response(user_text)