Collection of resources on the applications of Large Language Models (LLMs) in Audio AI.

Rapida is an open-source, end-to-end voice AI orchestration platform for building real-time conversational voice agents with audio streaming, STT, TTS, VAD, multi-channel integration, agent state management, and observability.

A New End-to-end Framework for Evaluating Voice Agents

Project for speech bubble

🇺🇦 Open Source Ukrainian Text-to-Speech datasets

A Docker-based OpenAI-compatible Text-to-Speech API server powered by Kyutai's TTS models with GPU acceleration support.

Just a simple multimodal avatar interaction platform

A voice-based AI chat interface built with Next.js and ElevenLabs. Start and stop real-time conversations with an animated UI that reflects agent status. Fully responsive and deployable via Vercel with environment-based agent configuration.

MLX Porting Toolkit — an agent-guided, evidence-gated pipeline (scaffold → convert → parity → benchmark) plus a portable skill for porting PyTorch/Hugging Face models to Apple MLX.

Open-source real-time Voice AI infrastructure in Go. Stream audio via WebRTC or WebSocket, connect STT → LLM → TTS pipelines, and build scalable voice agents and conversational AI applications.

A unified benchmarking framework for evaluating Voice AI agents across conversational quality, audio realism, latency metrics, and safety guardrails with scalable multi-language stress testing.

🇺🇦 Ukrainian RAD-TTS++ models (decoder + models with 3 voices) and HiFiGAN model

A voice-based AI chat interface built with Next.js and ElevenLabs. Start and stop real-time conversations with an animated UI that reflects agent status. Fully responsive and deployable via Vercel with environment-based agent configuration.

Production-ready examples for Brainiall Speech AI APIs — Pronunciation Assessment, STT, TTS. Python, JavaScript, curl, and MCP configs.

Code-switching ASR adaptation for strong multilingual speech recognition models. Synthetic CSW data generation, Whisper adaptation, Bayesian LoRA (BLoRA), and robust multilingual ASR evaluation.

MCP Server for Brainiall Speech AI - pronunciation assessment, speech-to-text, and text-to-speech

Enterprise-Grade Secure ASR Diarization Pipeline - HIPAA-compliant speech processing service combining automatic speech recognition with speaker diarization. Features modular architecture, comprehensive security, and production-ready deployment.

Open source AI voice calling agent for Twilio phone calls, built with FastAPI, Google ADK, and Gemini Live API

Enterprise voice operations platform for approved voice assets, transcription review, consent tracking, provider-cost governance, and OSS-ready deployment.

中文 ASR 评测工具箱 · micro-CER 对比 FunASR/Whisper/llama.cpp · 一条命令出报告 · 自带迷你测试集 · Mandarin ASR benchmark toolkit