EchoKit + Groq: Full-Stack AI Voice Solutions
EchoKit is a full-stack open-source AI solution written in Rust that provides everything from embedded firmware to AI agents to inference API servers for customizable AI voice experiences. Combined with Groq's blazing-fast inference, you can build responsive voice AI applications with minimal latency.
Key Features:
- ASR-LLM-TTS Pipeline: Modular architecture for speech recognition, language processing, and text-to-speech
- Cross-Platform: Works with embedded devices, mobile apps, and web applications
- OpenAI-Compatible: Seamless integration with Groq and other LLM providers
- Low Latency: Optimized for real-time voice interactions and streaming
- Customizable: Full control over system prompts, voice selection, and model parameters
- Production Ready: Built in Rust for performance and reliability
Quick Start
Follow the EchoKit Server Quick Start to install and run the server, then configure it with Groq models:
Configure your Groq API key and models:
curl
# Server settings
addr = "0.0.0.0:8080"
# Speech recognition (ASR)
[asr]
url = "https://api.groq.com/openai/v1/audio/transcriptions"
lang = "en"
api_key = "gsk_your_groq_api_key_here"
model = "whisper-large-v3-turbo"
# Language model (LLM)
[llm]
llm_chat_url = "https://api.groq.com/openai/v1/chat/completions"
api_key = "gsk_your_groq_api_key_here"
model = "openai/gpt-oss-20b"
history = 3
# Text-to-speech (TTS)
[tts]
platform = "Groq"
url = "https://api.groq.com/openai/v1/audio/speech"
api_key = "gsk_your_groq_api_key_here"
model = "playai-tts"
voice = "Aaliyah-PlayAI"
# System personality
[[llm.sys_prompts]]
role = "system"
content = """
You are EchoKit, a helpful AI assistant. Provide clear, concise responses and maintain a friendly, professional tone. Keep answers brief but informative.
"""Configuration Examples
Fast Performance Setup
Optimize for speed and responsiveness:
curl
# Optimized for speed
[asr]
model = "whisper-large-v3-turbo"
[llm]
model = "llama-3.1-8b-instant"
history = 1
[tts]
model = "playai-tts"
voice = "Cooper-PlayAI"
[[llm.sys_prompts]]
content = "Keep responses very concise, under 2 sentences. Be direct and helpful."Quality-Focused Setup
Prioritize response quality and context:
curl
# Optimized for quality
[asr]
model = "whisper-large-v3"
[llm]
model = "openai/gpt-oss-120b"
history = 5
[tts]
model = "playai-tts"
voice = "Sarah-PlayAI"