chat-with-anyone

The chat-with-anyone skill enables users to interact with real people or fictional characters through synthesized voice conversations. It works by locating publicly available speech samples online, extracting a clean reference segment, and generating spoken replies that imitate the target voice. The skill also supports generating a matching synthetic voice from an uploaded image when no recognizable speaker is identified. It is designed for immersive conversational experiences, roleplay interactions, narrated content, and voice-based storytelling while emphasizing ethical and responsible use.

The skill includes multiple workflows depending on user input. For named individuals or recognizable public figures, it searches online platforms such as YouTube or Bilibili for interviews, speeches, or press conferences with clear speech. It then downloads audio and subtitles, identifies the best reference segment automatically, and uses text-to-speech generation to synthesize responses in the target voice. For image-based workflows, it can create a matching voice profile from visual input. The skill also provides dependency verification steps for tools such as ffmpeg, yt-dlp, and the tts skill, along with setup instructions for API configuration.

This skill is intended for AI agents, developers, content creators, and experimental conversational applications that require realistic voice roleplay or narrated interactions. Common use cases include chatting with fictional characters, generating spoken article summaries in a celebrity-style voice, interactive storytelling, educational demonstrations, and creative entertainment experiences. The documentation also emphasizes safety requirements, including refusing deceptive impersonation, avoiding private individuals without consent, and informing users that generated speech is synthetic.