Scribd
Upload
21 views2 pages

Voice Assistant with LiveKit Plugins

This document outlines a Python script for a voice assistant application using LiveKit agents and various plugins like OpenAI and Deepgram. The assistant is designed to interact with users in a friendly, human-like manner, adopting different personas based on user requests and responding in Hindi. The script includes setup for voice activity detection, speech-to-text, language model, and text-to-speech functionalities.

Uploaded by

jagdishdhull13
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as TXT, PDF, TXT or read online on Scribd
21 views2 pages

Voice Assistant with LiveKit Plugins

This document outlines a Python script for a voice assistant application using LiveKit agents and various plugins like OpenAI and Deepgram. The assistant is designed to interact with users in a friendly, human-like manner, adopting different personas based on user requests and responding in Hindi. The script includes setup for voice activity detection, speech-to-text, language model, and text-to-speech functionalities.

Uploaded by

jagdishdhull13
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as TXT, PDF, TXT or read online on Scribd

import logging

from dotenv import load_dotenv


from [Link] import (
AutoSubscribe,
JobContext,
JobProcess,
WorkerOptions,
cli,
llm,
)
from [Link] import VoicePipelineAgent
from [Link] import openai, deepgram, silero

load_dotenv(dotenv_path=".[Link]")
logger = [Link]("voice-agent")

def prewarm(proc: JobProcess):


[Link]["vad"] = [Link]()

async def entrypoint(ctx: JobContext):


initial_ctx = [Link]().append(
role="system",
text=(
"You are a voice assistant created by Ajay. Your interface with users
will be voice. "
"You should respond quickly and use friendly, human-like conversation. "
"You will act like a friend, addressing people like 'brother', 'friend',
'mother', 'father', 'uncle', 'sister', etc., depending on the role requested by the
user. "
"If a user asks you to act like a specific character, such as 'brother' or
'father', you will immediately take on that persona and communicate as that
character, including using their tone, expressions, and mannerisms. "
"You will do your best to sound as human-like as possible, reflecting the
behavior of the character the user asks you to act like. "
"Always use short, concise, and polite responses in Hindi."
),
)

[Link](f"connecting to room {[Link]}")


await [Link](auto_subscribe=AutoSubscribe.AUDIO_ONLY)

# Wait for the first participant to connect


participant = await ctx.wait_for_participant()
[Link](f"starting voice assistant for participant {[Link]}")

# This project is configured to use Deepgram STT, OpenAI LLM and TTS plugins
# Other great providers exist like Cartesia and ElevenLabs
# Learn more and pick the best one for your app:
# [Link]
assistant = VoicePipelineAgent(
vad=[Link]["vad"],
stt=[Link](),
llm=[Link](model="gpt-4o-mini"),
tts=[Link](),
chat_ctx=initial_ctx,
)

[Link]([Link], participant)

# The agent should be polite and greet the user when it joins :)
await [Link]("Hey Ajay, how can I help you today?",
allow_interruptions=True)

if __name__ == "__main__":
cli.run_app(
WorkerOptions(
entrypoint_fnc=entrypoint,
prewarm_fnc=prewarm,
),
)

You might also like