Cloudflare + GPU voice workflow

VoiceForge

Upload permissioned audio, analyse style, generate audiobook MP3 chunks, store assets in R2, track jobs in D1, and call a separate GPU backend when serious inference is needed.

Checking stack…

Title Text

I confirm this text and selected voice workflow are original, open, or permissioned.

Generation settings

Preset GPU voice Speed 0.95x Optional reference asset

Boundary: this build supports permissioned voice workflows only. It does not clone celebrities, politicians, performers, friends, or family without consent.

Upload permissioned audio

Supports MP3, M4A, WAV, AAC, OGG, and FLAC. Files go to Cloudflare R2; metadata goes to D1.

Owner / rights note Choose audio file I confirm this uploaded voice/audio is mine, open-licensed, or explicitly permissioned.

Recent jobs

Deployment settings that fix your error

Your previous Cloudflare build failed because the project was deployed as if it were Next.js. This project is not Next.js.

Framework presetNone

Build commandnpm run build

Output directorydist

Install commandnpm install

Then bind D1 as DB, R2 as AUDIO_BUCKET, and set GPU_BACKEND_URL.