ElevenLabs — многоголосая озвучка диалогов

★ 7.7 · content

elevenlabs-dialogue is a Claude Code skill that generates multi-speaker dialogue as a single audio file via ElevenLabs and the inference.sh belt CLI. It takes a JSON script where each segment specifies the line of text and the assigned voice, then runs `belt app run elevenlabs/text-to-dialogue` to produce the output. Over 22 premium voices are available — 8 female (aria, alice, bella, jessica, laura, lily, sarah, matilda) and 14 male — and delivery can be shaped with bracket-style direction tags such as [excitedly], [whispering], or [sarcastically]. Ideal for podcasts, audiobooks, explainer videos, product demos, and customer-service training scripts where multi-voice audio is needed without a recording studio.