Real-time AI voice announcer for races, benchmarks, and long-running agent processes in Claude Code. Watches your logs, generates sports-style play-by-play commentary, and speaks it aloud using fast local text-to-speech (TTS). Powered by Claude. No cloud TTS APIs required.

No description available

Latency benchmark for real-time speech-to-text APIs: Groq, OpenAI, Azure, Deepgram, Together, Google Gemini

AI Ground Truth Audio Generator - Synthetic Voice Dataset Creation with ElevenLabs API This repository contains a high-performance Python utility designed to automate the creation of high-quality audio datasets. It is specifically engineered to generate "Ground Truth" audio files for testing and benchmarking Speech-to-Text (STT) systems.

SpeechEvalAI is a comparative analysis project that evaluates the performance of state-of-the-art speech-to-text models, specifically OpenAI Whisper API and Facebook Wav2Vec2. The project benchmarks models based on transcription accuracy, error metrics, and real-world audio datasets, providing insights into their strengths and limitations.