VoiceLabs vs Fish Audio
Fish Audio’s plan names, credits, minute figures and free-tier inclusions below were read from fish.audio/plan as of 2026-09-02. The minute figures on their rows are theirs, not ours — their page publishes minutes alongside credits. Their plans change; check their site for the current list. VoiceLabs prices come straight from our own plan configuration.
| DIMENSION | VoiceLabs | Fish Audio |
|---|---|---|
| Availability | Live. The studio, the seven TTS engines, instant voice cloning, downloads and captions all run today, as do accounts, the Pro plan with billing, the REST API and our MCP server for AI agents. | Shipped and generally available — you can sign up and generate today. |
| Free plan | There is no free PLAN and no guest mode: VoiceLabs Pro is the only thing you can buy, and you sign in to generate. | US$0/month with 8,000 credits monthly, up to 500 characters per generation — their own page puts that at 7 minutes of generation — plus 3 public voice slots. |
| Cloning and commercial use on the free plan | Commercial rights come with VoiceLabs Pro; a free account is for personal use. That is the same gate ElevenLabs, Murf, Speechify Studio and Cartesia use. | Both included. Their free plan card lists enhanced voice cloning and commercial use, which makes it the most generous free tier of any vendor we benchmarked. This is a genuine advantage over us and over most of the market. |
| Monthly volume for the money | 2 hours of generated audio a month as fair use, for US$8/month or US$59/year (US$4.92/mo billed annually), with a 7-day trial. | Plus is US$11/month for 250,000 credits per month — up to 200 minutes, their figure. |
| Per-generation length limit | No fixed per-request character cap on the plan. Long scripts are split and stitched, and the splitter prices each chunk by writing system so a Hindi or Chinese script is not silently truncated the way a character-only limit truncates it. | Their free plan is listed as up to 500 characters per generation, so long-form work on the free tier means many separate requests. |
| Engines you can choose | Seven TTS engines — Qwen3-TTS, Chatterbox, Kokoro and others — picked per generation, with 23 languages across them. | Their own models, with a community voice library on top. We have not verified their engine or language list, so we make no claim about it here. |
| AI agent access (MCP) | A remote MCP server, live now: streamable HTTP, OAuth 2.1 with a consent screen, seven tools, revocable per agent. Nothing to install and no API key to paste. | We did not verify whether Fish Audio publishes an MCP server, so we make no claim either way — check with them. |
| Open-source foundation | Built on the open-source VoiceBox project (MIT), and that attribution is permanent. The hosted service around it is commercial. | They publish open-source model work of their own. We have not verified the licensing of what powers their hosted product, so we make no claim about it here. |
Who should pick Fish Audio
Anyone who wants to clone a voice and use it commercially without paying anything. That is a real offer and it is theirs alone in this comparison set — ElevenLabs, Murf, Speechify Studio and Cartesia all charge for at least one of those two things, and so do we. If you are experimenting, if your project may never earn anything, or if you simply want to hear what a clone of your own voice sounds like before spending money, start there.
Who might prefer VoiceLabs
People whose scripts are longer than a paragraph and whose bill needs to be predictable. Fish’s free tier is generous but capped at 500 characters per generation and 7 minutes a month; past that you are on a credit plan again. VoiceLabs is one flat US$8 a month for 2 hours of audio with no per-character fee, seven engines to choose between per generation, a text splitter that accounts for dense writing systems instead of cutting them off, an MIT open-source foundation, and a hosted MCP server so an AI agent can generate through an OAuth consent screen. Their free plan, in the table above, is the thing to weigh against it.
Questions about VoiceLabs and Fish Audio
- Is VoiceLabs cheaper than Fish Audio?
- On the monthly bill, yes: VoiceLabs Pro is US$8 a month, and Fish Audio's cheapest paid plan, Plus, is US$11 a month (fish.audio/plan, read 2026-09-02). Plus includes more audio, though: up to 200 minutes a month by their figure, against 2 hours on VoiceLabs. Fish Audio also has a US$0 plan.
- Can I clone a voice?
- Yes — instant voice cloning from a short reference clip, in seconds. You may only clone a voice from a sample you are authorized to use, and our Responsible Use policy applies to every generated voice. Fish Audio includes cloning even on its free plan: the plan card lists enhanced voice cloning and commercial use (fish.audio/plan, read 2026-09-02).Responsible Use policy
- Does VoiceLabs have an API and an MCP server?
- Yes. VoiceLabs runs a remote MCP (Model Context Protocol) server, so an AI agent can work with your VoiceLabs data on your behalf after you sign in and consent. It uses OAuth 2.1, exposes seven tools, and you can revoke any agent's access at any time. VoiceLabs Pro also includes a REST API at /v1, with an OpenAI-compatible speech endpoint. We did not verify whether Fish Audio publishes an MCP server, so we make no claim either way.MCP server docsDeveloper API docs
- How do I move from Fish Audio to VoiceLabs?
- Bring your scripts and the original recording of any voice you cloned on Fish Audio, then clone it again here: no platform, VoiceLabs included, can import another vendor's clone. VoiceLabs does not carry Fish Audio's stock voices, so anything narrated by one needs a new voice. If your code calls Fish Audio, its requests need rewriting for the VoiceLabs REST API at /v1.Developer API docs
7-day free trial, card upfront, cancel anytime before it ends.
Fish Audio is a trademark of its respective owner. VoiceLabs is not affiliated with, endorsed by, or sponsored by Fish Audio. See our changelog for what has actually shipped on our side.