#RealtimeTTS
ブラウザベースのリアルタイムAI音声対話の実装。
RealtimeSTT(20ms単位でWebSocketで送信) → LLM → RealtimeTTS のそれぞれのモジュールがカスタマイズできて、割り込みなども実装されているっぽい?
これWhisperとLLM箇所をOpenAIのバックエンド叩いてもリアルタイム性担保できるのかな?GPU非依存で外部サービスつなぎこむことで、どこまでできるか試してみたい。そもそも日本語いけるのかも含めて
github.com/KoljaB/Realt...
GitHub - KoljaB/RealtimeVoiceChat: Have a natural, spoken conversation with AI!
Have a natural, spoken conversation with AI! Contribute to KoljaB/RealtimeVoiceChat development by creating an account on GitHub.
github.com
May 6, 2025 at 2:15 PM
March 23, 2026 at 6:30 PM
September 1, 2025 at 8:31 PM
🚀 Inworld AI just dropped Realtime TTS‑2, now with three stability modes to keep voices smooth in any scenario. Curious how this changes virtual chats? Dive in! #InworldAI #RealtimeTTS #StabilityModes

🔗 aidailypost.com/news/inworld...
May 6, 2026 at 2:57 AM
Theme 2: System Architecture. The project uses WebSockets for audio, RealtimeSTT (Whisper), & RealtimeTTS. Discussion covered performance of components & hardware needs (CUDA GPUs). Users explored STT/TTS alternatives. #VoiceAI 3/6
May 6, 2025 at 2:21 PM