PLOPS FORGE · VOICE STUDIO · OCTOBER 25, 2026

Ming 3.0 text-to-speech on Plops Forge

Create a long voiceover with Ming 3.0: choose a voice, direct its acting, and download one WAV.

Open Ming voice studio

Write your script, pick a voice, and tell it how to perform. Ming 3.0 speech generation is now available on Plops Forge using prepaid credits.

Your voice and acting direction, together

James and Stella are built in. You can also upload an authorized reference recording, supply its exact transcript, and save it in this browser. Choose all of these voices from the same picker. Add a direction such as “warm and reassuring, with a measured pace.” Combined voice conditioning and acting direction are experimental, so listen to a short take before committing to a long one.

Long scripts become one recording

The studio has no 1,200-character script-field limit. It divides text into chunks of up to 500 characters, keeps the voice and direction consistent across requests, and combines the completed chunks into a single WAV. Available credits, compute allowance, request size, and storage still apply.

Know the price before you start

The current rate is 100 credits per chunk. The studio shows the number of chunks and total credits before submission. A failed take returns its generation credits. Credit packs do not reserve compute capacity, and generation pauses when the shared allowance is exhausted.

Follow progress from another tab

The browser title shows the percentage of completed chunks. When every chunk is joined, it reads “100% Done.” Takes and downloadable audio are private to the generating account.

What to expect

Workers shut down when idle, keeping hosting costs low. The first request can take longer while a worker starts. Forge currently uses Ming 3.0’s 0.5B variant and generates one speaker per take. Text and selected reference audio are sent to the RunPod worker.

Model source and documentation: inclusionAI/Ming-3.0-tts. The upstream source is MIT licensed.