wtf is voxweaver-studio?

VoxWeaver Studio is a web application you run on your own computer for producing AI generated voice audio. It combines a text to speech service called Fish Audio with a language model called DeepSeek to optimize written text before turning it into speech, and it can handle everything from a single piece of text to large multi role scripts. For a single script, you can type or upload a text file, optionally let DeepSeek clean up or improve the wording, add emotion and style cues, choose a voice, and generate a preview before downloading the final audio file with a timestamped name. For larger jobs, a batch mode accepts multiple text files at once, applies AI optimization to each one individually, and exports all the finished audio as a single zip file. The Studio section is built for scripts with several characters. You paste in a long piece of text, ask the AI to split it into roles, review or manually fix any parts it could not confidently assign, and then have each role audition different voices, generate a full preview, and lock in a voice that gets saved as a reusable clone for that character. Finished lines then queue up and get merged into one combined output file. Voices can come from a default library, an existing Fish Audio voice referenced by its ID, an instant clone created from short reference audio, or a persistent clone that Fish Audio trains and stores under your account. Running it locally requires Node.js 18 or newer, the pnpm package manager, and ffmpeg installed and available on your system path, along with API keys for Fish Audio and DeepSeek stored in a local .env file. The interface can be switched between Chinese and English from the main screen. The README stresses that API keys are read only on the backend server and should never be placed in frontend code or committed to version control, and it includes a troubleshooting table for common setup errors.