AI voice generation with TTS models

★ 7.7 · content

ai-voice-cloning is a Claude Code skill that generates natural-sounding audio via the inference.sh CLI (`belt`), supporting models including Inworld TTS-2, Inworld TTS 1.5 Mini/Max, ElevenLabs, Kokoro, DIA, Chatterbox, Higgs, and VibeVoice. Inworld TTS-2 covers 100+ languages with 271+ built-in voices and inline emotion steering tags like `[excited]`, `[whisper]`, and `[gasp]`; the 1.5 Mini variant achieves ~120 ms latency for real-time conversational AI; ElevenLabs delivers premium quality across 32 languages with a voice changer for transforming existing recordings. Use cases range from video voiceovers, audiobooks, and podcasts to gaming NPC voices, avatar audio, and accessibility. It fits developers and content creators who need realistic speech synthesis or expressive character voices directly from Claude Code.