Text to Speech MCP Server — Hosted TTS for AI Agents (2026)
Every TTS MCP server in the directories is a local stdio process — install Python, set an API-key variable, meter your characters. EasyVoice’s is a hosted URL: paste it into Claude, Cursor, or ChatGPT developer mode and your agent gets the whole platform — instant speech, long-form narration, two-speaker podcast episodes, and voice cloning — over one endpoint, with your existing API key.
A free key works today for instant speech — 12 voices, no card. Pro is $9.99/month flat with unlimited characters: no per-character meter, so agent workflows have a fixed bill.
Free key: 12 voices, 5,000 characters/day, no card. Pro $9.99/mo flat unlocks all 56 voices plus cloning, podcast, and long-form over the same URL — 14-day money-back guarantee.
Verified July 2026 · Last updated July 21, 2026. Tool names, limits, and pricing checked against the shipped EasyVoice MCP server and public competitor docs before publication.
The TTS MCP server in four facts
- EasyVoice's MCP server is hosted — paste https://easyvoice.ae/api/mcp into Claude, Cursor, or ChatGPT developer mode and generate speech with a free API key: no local install, no card. It's the only flat-rate hosted TTS MCP server we could find as of July 2026.
- EasyVoice's MCP server exposes the whole platform to AI agents — text to speech in 56 voices, long-form narration up to 500,000 characters (~9 hours of audio) per job, two-speaker podcast episodes, and consent-verified voice cloning. It's the only MCP server we could find that offers voice cloning and podcast generation hosted at a flat rate (July 2026). Cloning, podcast, and long-form require Pro ($9.99/month flat).
- A free EasyVoice API key gives any MCP agent 12 voices at 20 requests/minute, 2 concurrent requests, and 5,000 characters/day shared with the web app — no card required.
- Any MCP-capable AI agent can generate speech from one hosted URL — $9.99/month flat, unlimited characters (10M/month fair use), so an agent loop can't surprise-bill you.
What is a TTS MCP server?
MCP (Model Context Protocol) is how AI assistants gain abilities: you point the assistant at a server URL, and every tool that server exposes becomes something it can do on its own. A TTS MCP server exposes speech tools — an API the agent drives itself, no SDK, no copy-pasting between apps.
EasyVoice’s server exposes seven tools covering the full platform, in four groups:
- Generate —
text_to_speech: instant speech, returned as a hosted MP3/WAV download link. This is the tool free keys can use. - Produce —
generate_long_form,create_podcast, andget_job_status(Pro): the agent submits a job, gets a job id back, and polls status until the audio URL appears — queued, then polled, never a long blocking call. - Voice —
clone_voice(Pro, consent-verified enrollment) andlist_voices(the 56-voice catalog plus your clones and their status). - Account —
get_usage: plan, remaining free characters, and rate limits, so agents can handle quota errors instead of retrying blindly.
Audio always comes back as a download URL, not inline bytes — links any client, CI job, or teammate can open. Download promptly: links are not guaranteed to persist.
Hosted remote vs local stdio — how TTS MCP servers compare (verified july 2026)
The short version: every alternative runs on your machine and meters your usage (or bills you for the GPU). A hosted flat-rate server is the only shape that works from claude.ai in a browser, a phone-configured connector, or a CI runner — with a bill that doesn’t scale with the agent’s enthusiasm.
| Server | How it runs | Pricing | What the tools expose |
|---|---|---|---|
| EasyVoice MCP (hosted remote) | Paste one URL into your client — no install, no local process | Free API key today (no card); Pro $9.99/mo flat, unlimited (10M chars/mo fair use) | Text to speech, long-form narration, podcast episodes, voice cloning, job status, voice listing, usage — over one URL; Pro unlocks cloning/podcast/long-form |
| ElevenLabs official MCP | Local stdio — uvx/pip install, API-key environment variable | Metered credits: free 10K credits/mo, then $5–$22+/mo tiers | TTS, voice cloning, speech-to-text, sound effects — honest note: it does expose cloning; output is saved to the local Desktop by default |
| MiniMax-MCP (official) | Local stdio — install + API-key environment variable | Metered API billing | TTS-family tools returning local file paths — same install-and-meter pattern |
| Community wrappers / self-hosted Kokoro | Local stdio — your own machine, GPU, or self-hosted endpoint | Your own GPU or API bill | Typically TTS-only |
Rows re-verified against public READMEs and pricing pages, July 2026. Honest note: if you want a local stdio server and already pay for metered credits, the ElevenLabs MCP works well — and it exposes cloning tools too. The hosted, flat-rate packaging is what this page is about, and both of our “only” claims above keep the “we could find” hedge for exactly that reason.
What agents actually do with it
Each of these is exactly what the shipped tools do — input in, job or URL out:
- Voice notifications from agent workflows — an agent calls
text_to_speechon the text it just wrote and posts the audio link: spoken build results, narrated summaries, alert readouts. - Long-form narration pipelines (Pro) — the agent submits a script of up to 500,000 characters with
generate_long_formand polls until the finished MP3 URL is ready. Kokoro narration voices only. - Agent-drafted podcast episodes (Pro) — the agent writes the dialogue, calls
create_podcastwith up to 60 speaker A/B segments (30,000 characters per episode), and collects the automatically stitched episode. - Clone once, podcast in your voice (Pro) — enroll a consent-verified clone with
clone_voice, then use the READYvoice_*id as speaker A or B in agent-drafted episodes. Plain narration in a cloned voice is not an MCP capability —text_to_speechrejectsvoice_*ids and long-form is Kokoro-only; that lives onPOST /api/v1/audio/speechand the web studio. - CI and automation jobs that speak — anything that can bridge MCP (or call the API) can emit spoken status without owning a GPU or a per-character budget.
Connect in 2 minutes
- Create a free account (no card) and generate an API key at /account/keys.
- Copy the server URL:
https://easyvoice.ae/api/mcp. - Paste the endpoint (or the one-liner below) into your client.
In Claude Code, it’s a single command:
claude mcp add --transport http easyvoice https://easyvoice.ae/api/mcp --header "Authorization: Bearer ev_your_api_key"
Per-client instructions — claude.ai connectors (API-key headers are beta there), ChatGPT developer mode, Cursor, and the mcp-remote bridge for stdio-only clients — live on the MCP docs page, and easyvoice-mcp.md is a downloadable connection file you can hand to any AI assistant.
TTS MCP server — frequently asked questions
What is an MCP server for text to speech?▾
MCP (Model Context Protocol) is the standard AI assistants use to gain abilities: you give the assistant a server URL, and every tool that server exposes becomes something it can do. A TTS MCP server exposes speech tools — connect one and Claude, ChatGPT, Cursor, or any MCP-capable agent can turn the text it writes into audio. EasyVoice's is hosted: paste https://easyvoice.ae/api/mcp into your client with an API key, and audio comes back as hosted MP3/WAV download links — download promptly, links are not guaranteed to persist.
How do I connect Claude to the EasyVoice MCP server?▾
On claude.ai and Claude Desktop: Settings → Connectors → add a custom connector with the URL https://easyvoice.ae/api/mcp, then add the request header Authorization: Bearer ev_your_api_key (the API-key header option is labeled beta in the connector UI). In Claude Code it's one line: claude mcp add --transport http easyvoice https://easyvoice.ae/api/mcp --header "Authorization: Bearer ev_your_api_key". Keys are free at easyvoice.ae/account/keys — no card.
Does it work with ChatGPT and Cursor?▾
ChatGPT developer mode (web) accepts remote MCP servers, but its connector form has no custom-header field — so connect with the key-in-URL form https://easyvoice.ae/api/mcp?key=ev_your_api_key and authentication set to none. One hedge: community reports say ChatGPT Plus and Pro may restrict custom-connector tools to read-only, which would block text_to_speech there — this is multiply reported but not confirmed in OpenAI's docs, so treat write tools on Plus/Pro as unverified. Cursor connects via ~/.cursor/mcp.json with a url plus a headers entry carrying the Authorization header. Any stdio-only client can bridge with npx mcp-remote.
How do I make Claude generate voice audio?▾
Claude can't synthesize speech by itself — you connect it to a text-to-speech service over MCP. Add EasyVoice's server as a custom connector (or via claude mcp add in Claude Code) with a free API key, then ask Claude to call the text_to_speech tool: it returns a hosted MP3 or WAV download link. A free key covers 12 voices and 5,000 characters/day, no card.
Can an AI agent produce a whole podcast episode?▾
Yes, with a Pro key. The create_podcast tool takes alternating speaker A/B dialogue segments — up to 60 segments and 30,000 characters per episode — with a voice for each speaker. The job is queued, the agent polls get_job_status, and the completed job carries the stitched episode URL plus per-segment audio URLs. Free keys get a truthful Pro-required error instead.
Can an AI agent clone a voice over MCP?▾
Yes — enrollment, with guardrails, on a Pro key. The clone_voice tool requires a consent: true attestation (recorded server-side) and an audio sample passed as exactly one of audio_base64 or a public https audio_url — 10 MB max, wav/mp3/m4a, a 15–60 second clean recording works best. Account limits: 3 cloned voices maximum, 5 enrollments per day; cloned output carries an AudioSeal watermark. Using the clone over MCP is podcast-only: create_podcast accepts your READY voice_* ids as speaker A or B, while text_to_speech rejects voice_* ids and generate_long_form is Kokoro-only — synchronous narration in a cloned voice lives on POST /api/v1/audio/speech and the web studio, not the MCP tools.
Does long-form narration work from an agent?▾
Yes, with a Pro key. The generate_long_form tool accepts a script up to 500,000 characters (~9 hours of audio) in a single job — Kokoro narration voices only; Arabic ar_* and cloned voice_* voices are rejected by the long-form engine. The agent polls get_job_status: parts render progressively (N of M), and the download URL appears when the job completes.
What happens if I call a Pro tool with a free key?▾
A truthful error, nothing else. The job tools return the Jobs API's own message — "The Jobs API requires a Pro subscription — free API keys can use /v1/audio/speech." — as a tool error, and clone_voice returns its own Pro-required message. Nothing is queued and nothing is charged. Pro is $9.99/month flat; there are no promo codes to hunt for.
What does it cost — and is there a free tier?▾
A free API key works today for sync text_to_speech: 12 voices, 20 requests/minute, 2 concurrent requests, and 5,000 characters/day shared with the web app — no card required. Pro is $9.99/month flat: all 56 voices, unlimited characters (10M/month fair use), and the cloning, podcast, and long-form tools over the same URL — with a 14-day money-back guarantee.
How is it different from the ElevenLabs MCP server?▾
ElevenLabs' official MCP server is a local stdio process: you install it with uvx or pip, set an API-key environment variable, and metered credits count every character; its tools include TTS, voice cloning, speech-to-text, and sound effects, with output saved to the local Desktop by default. EasyVoice's MCP server is hosted — a URL, not an install — and flat-rate at $9.99/month unlimited. To be precise about the overlap: ElevenLabs does expose cloning tools too; the difference is the hosted, flat-rate packaging, and it's the only MCP server we could find offering voice cloning and podcast generation hosted at a flat rate as of July 2026.
Can AI agents use it for voice notifications and pipelines?▾
Yes — that's the core use case. Any MCP-capable agent or automation can call text_to_speech and get a hosted audio URL back: CI jobs that announce deploy status out loud, agent workflows that voice their summaries, apps that need spoken alerts. Because Pro is $9.99/month flat with unlimited characters (10M/month fair use), an agent calling the tool in a loop can't surprise-bill you.
What tools does the server expose?▾
Seven: text_to_speech (synchronous speech — up to 8,000 characters per call, 4,000 on Arabic ar_* voices), list_voices, get_usage, generate_long_form (Pro), create_podcast (Pro), clone_voice (Pro), and get_job_status. Every tool returns a plain-text summary plus structured results, and audio always arrives as a hosted download URL rather than inline bytes.
Free key: 12 voices, 5,000 characters/day, no card. Pro $9.99/mo flat unlocks all 56 voices plus cloning, podcast, and long-form over the same URL — 14-day money-back guarantee.