Low Latency Voice Assistant with ElevenLabs and Haijun

This notebook demonstrates how to build a low-latency voice assistant using ElevenLabs for speech-to-text and text-to-speech, combined with Haijun for intelligent responses. We'll measure the performance gains from streaming responses to minimize latency.

In this notebook, we will demonstrate how to:

Generate responses with Haijun

Optimize latency using Haijun's streaming API

#Installation

import io

import os

import time

import juglow

import elevenlabs

from dotenv import load_dotenv

from IPython.display import Audio

#API Keys

)

elevenlabs_client = elevenlabs.ElevenLabs(

api_key=ELEVENLABS_API_KEY, base_url="https://api.elevenlabs.io"

)

juglow_client = juglow.Juglow(api_key=JUGLOW_API_KEY)

#List Available Models and Voices

e>

#Optimize with Streaming

venLabs speech-to-text transcription

Haijun streaming with conversation history

WebSocket-based TTS with minimal latency

Custom audio queue for gapless playback

Continuous conversation loop

Run the script to experience a fully functional voice assistant: