From your first words
to your next project.

Your first voiceover

1. On the homepage, choose a featured voice and press its play button to hear a recording. Try an example script or enter up to 120 characters, then generate a free preview.
2. Create an account to open the full workspace. New accounts include 2,000 welcome credits. Your demo draft and voice choice can carry over in the same browser when local storage is available.
3. In Create speech, choose a library or saved voice, enter up to 5,000 characters and review the credit estimate. Submit the generation once.
4. Follow the job status. When the audio is ready, listen and download it. Recent generations keeps your latest 30 signed-in jobs, including API and MCP jobs.

Sample playback is an existing recording. Generating your own text creates a new result. Listening, seeking and changing playback settings do not regenerate audio or spend generation credits.

Clone a sample or save a voice

For one-off cloning, choose the clone option and supply a reference recording. For a voice you want to use again, open My voices.

1. Choose a clear recording of one speaker, with little background noise. Use audio you own or have permission to use. Keep it under 4 MB and about 59 seconds; the server rejects audio longer than 60 seconds.
2. Give the voice a name and upload the reference. Enter its transcript or use Generate transcript, then review the words for accuracy.
3. Review the 10,000-credit saving cost before saving. This keeps the reference and transcript for reuse; it does not train a separate model.
4. Select the saved voice in Create speech. Integrations can also find it through the available voice list. Generating with it uses five credits per normalized UTF-8 byte.

The 2,000 welcome credits can help you try speech generation, but do not cover the 10,000-credit saved-voice fee. The public cloning preview uses a fixed sample sentence.

Work in paragraphs. Export together.

1. Open Projects, give your project a name and enter its script. Creating the project saves the text; it does not generate audio.
2. Select a voice and work through the paragraphs. After editing, use Save changes before generating that paragraph.
3. Listen to each result. An edit makes that paragraph's earlier recording stale, so generate it again before exporting.
4. When every paragraph has current saved audio and there are no unsaved edits, choose Merge and download project.

Unsubmitted drafts live in the current browser. Saved projects belong to your account. If a newer save conflicts with yours, review the conflict before trying again. Keep a copy of important unfinished text before clearing browser data.

Credits, waiting and recovering your work

Check the estimate first

Library speech uses one credit per normalized UTF-8 byte. Saved custom voices and one-off cloning use five. A character can contain more than one byte, so the estimate can differ from the visible character count. Current packages are one-time credit purchases.

Choose a package or your own amount

• Starter: $5 for 100,000 credits. Cost per 1,000 ASCII characters with full-package use: $0.05 for library speech, or $0.25 for custom/reference speech.
• Creator: $35 for 1,750,000 credits. Cost per 1,000 ASCII characters with full-package use: $0.02 for library speech, or $0.10 for custom/reference speech.
• Premium: $100 for 10,000,000 credits. Cost per 1,000 ASCII characters with full-package use: $0.01 for library speech, or $0.05 for custom/reference speech.

All packages include the same tools. A custom top-up starts at $10 for 300,000 credits. Its rate rises smoothly from 30,000 credits per dollar at $10 to 100,000 per dollar at $100, then stays at that rate for larger purchases. The maximum is $10,000 per purchase. Review the credit quote and checkout total before paying; a successful return from checkout alone does not confirm a settled payment.

Follow the existing job

Accepted jobs reserve credits and may wait for available processing capacity or model startup. A terminal failure or cancellation while queued returns the reservation once. Work that has started cannot currently be cancelled. If your connection drops, recover the existing job instead of submitting the same speech as a new request.

Reopen saved audio for a fresh link

Customer audio is private. A signed download link lasts 15 minutes; reopen the saved generation to obtain a fresh one. Link expiry is separate from audio retention. Treat the signed link as private while it is valid.

Your voices, in your apps and AI tools.

The REST API and MCP use the same account credits, available voices and generation history as the workspace. Start in Developers to manage keys and connections.

For your app: REST API

Create a scoped key with an expiry and credit budget. Keep its secret on your server. List voices, estimate the cost, submit with an idempotency key, then check the returned job until it finishes. Poll only active jobs and stop at completion or after a bounded wait.

For a compatible AI client: MCP

Connect to https://www.vocalcopycat.com/mcp. Review permissions and the spending budget before authorizing. The current tools list voices, estimate and create speech, check a job, and read your balance. OAuth access expires after one hour; authorize again when needed.

An integration budget limits access to your existing credits; it does not add a separate balance. You can revoke keys and connections in Developers. Saved-voice creation remains a browser workflow.