Run your audio from your AI tools. Or call the voice directly.
The Everlit MCP server plugs Claude, Cursor, Claude Code and any MCP agent into your publications: publish an article from a URL, curate a playlist, switch on a podcast feed, fix a pronunciation, pull the numbers. The TTS API is the same Everlit Voice AI model on its own, for when you need the audio and nothing else.
Studio customers connect with OAuth today, no key to copy. Everyone else: request access and most hear back within a business day. Read the docs.
# Connect once. You log in to Studio; there is no key to copy.
claude mcp add --transport http everlit https://api.everlit.audio/v1/mcp
# Then just ask:
> Publish https://example.com/news/city-budget to The Daily,
> pin it to the Morning Briefing playlist, and send me the player link.
# list_publications → create_article → get_article → add_to_playlist
# ✓ https://everlit.audio/embeds/artl_9f3k... (4:14, Penny, intro bed 1201)- MCP tools
- 52
- connect, no key to copy
- OAuth
- languages
- 69
- per million chars, TTS API
- $8
Connect your agent, or take a key
Both are built on the Everlit Voice AI model that narrates for newsrooms and publishers every day. They solve different problems: the MCP server runs your Everlit account from whatever AI tool you already use; the TTS API puts the voice inside your own product.
Agentic Interface (MCP)
Claude, Cursor, Claude Code, VS Code and any MCP client
# Settings → Connectors → Add custom connector
Name: Everlit
URL: https://api.everlit.audio/v1/mcp
# Click Connect. You log in to Everlit Studio, pick the publisher
# the connector may act as, and approve. No key is ever shown.Then talk to it. “Everything we published this week that isn’t in a playlist yet: add it to Weekend Reads, then give me the RSS link.” The agent lists, filters, pins, and hands back the feed. Nothing to script.
TTS API
Any language with an HTTP client, or the OpenAI SDK
import os, time, requests
BASE = "https://api.everlit.audio"
H = {"Authorization": f"Bearer {os.environ['EVERLIT_API_KEY']}"}
job = requests.post(f"{BASE}/v1/speech", headers=H,
json={"text": "Hello from Everlit.", "voice": "669"}).json()
while job["status"] in ("queued", "processing"):
time.sleep(3)
job = requests.get(f"{BASE}/v1/speech/{job['id']}", headers=H).json()
with open("hello.mp3", "wb") as f:
f.write(requests.get(job["audio_url"], headers=H).content)Send text, get back a job with a stable audio URL, and pick it up by polling or webhook. MP3, WAV or Opus; word timings and captions included. Already on the OpenAI speech endpoint? Change the base URL and keep your code.
Your whole audio operation, from a prompt
Everlit puts a full audio operation behind whatever AI tool or workflow you already use. Our MCP server is the door in, and it is not a wrapper around text-to-speech: it exposes the things you actually do in Studio. Publish and edit articles from any source, set publication defaults, build playlists, run podcast feeds, keep a pronunciation library, manage music beds and voices, and read usage. Chain them and you have automation that used to need a developer.
Connect with OAuth, no key to copy
Add the URL, log in to Studio, pick a publisher, approve. Static keys remain for CI and headless agents.
Workflows, not one-shots
Publish → pin to a playlist → activate the feed → report the numbers. Every step is a tool, so the agent runs the whole chain.
Standing orders, too
Point a publication at an RSS or Atom feed and every new item is narrated on a schedule. Set it once from the conversation and forget it.
Every article, whatever created it
The widget, WordPress, RSS, uploads, the API: one list, one id scheme, one way to edit or take down.
Safe by annotation
Read-only and destructive tools are flagged, so your client confirms a delete and never confirms a list.
Docs and workflows on board
Both API guides and the language catalog ship as MCP resources, plus narrate_text and publish_from_url prompts for the slash menu.
Behaves in production
Running jobs say how long to wait, errors carry the same structured envelope as REST with a retry-after, and every call is metered exactly like Studio.
No Everlit account yet? The server offers request_access to a client with no credentials, so an agent can sign its user up from inside the conversation.
52 tools · 38 publishing, 11 speech & voices, 3 account
https://api.everlit.audio/v1/mcpArticles
Publish, edit, list and take down audio articles from any source, with pauses, music and second voices inline.
- create_article
- get_article
- list_articles
- update_article
- delete_article
Publications
Read and change the defaults every future article inherits.
- list_publications
- get_publication
- update_publication
Playlists
Curate tracks, set fill rules, reorder, and embed.
- list_playlists
- get_playlist
- create_playlist
- add_to_playlist
- remove_from_playlist
- move_in_playlist
- update_playlist
- delete_playlist
Podcast feeds
Fill show metadata, switch distribution on, and track where the show stands on each directory.
- get_feed
- update_feed
- activate_feed
- deactivate_feed
- list_feed_platforms
- mark_feed_submitted
- mark_feed_live
- reset_feed_platform
- ping_feed
RSS ingestion
Point a publication at an RSS or Atom feed and every new item is narrated.
- list_ingestion_feeds
- create_ingestion_feed
- get_ingestion_feed
- update_ingestion_feed
- delete_ingestion_feed
- poll_ingestion_feed
- list_ingestion_items
Pronunciations
Fix how a name is read, account-wide or per publication.
- list_pronunciations
- set_pronunciation
- delete_pronunciation
Music beds
Your intros, outros and stings, plus Everlit's library.
- list_mixables
- create_mixable
- delete_mixable
Speech
Text to a finished file, with timings, captions and estimates.
- create_speech
- estimate_speech
- get_speech
- list_speech
- get_speech_timings
- get_speech_captions
- delete_speech
Voices
Browse the catalog, clone a voice, check or remove a clone.
- list_voices
- create_voice
- get_voice
- delete_voice
Account
Who am I, what can I do, what has it cost.
- get_account
- get_usage
- list_languages
Articles count against your publication’s monthly plan; speech is included on every paid Studio plan and pay as you go otherwise. Nothing is metered differently because an agent asked. Articles & publishing reference · MCP reference
Things people ask it to do
Each of these is one message. The agent picks the tools; you approve the destructive ones.
Morning publish run
“Publish these five URLs to The Daily, pin them to Morning Briefing in this order, and give me the embed for the lead story.”
- create_article
- get_article
- add_to_playlist
- move_in_playlist
Archive to podcast
“Make a playlist of every article tagged "climate", fill it automatically from now on, set the show metadata from our About page, turn the feed on, and give me the Apple and Spotify submission links.”
- create_playlist
- update_feed
- activate_feed
- list_feed_platforms
Newsroom housekeeping
“Narrate everything our partner's RSS feed publishes from now on into the Wire publication, add these 30 surnames to the pronunciation library, and tell me what we spent on speech this month.”
- create_ingestion_feed
- set_pronunciation
- get_usage
One number. No plan required.
Prepaid and independent of any Everlit subscription. Add a card, buy credit, and every successful job is deducted at the published rate—in exact fractions of a cent. On a paid Studio plan, speech is already included: no per-character charge, no cap. Connecting the MCP server costs nothing; what you do through it is billed exactly as it would be in Studio.
For Latin, Cyrillic, and Greek scripts. Other scripts are weighted so every language costs about the same per finished hour of audio—see below.
- Word-level timestamps and captions included
- MP3, WAV or Opus at the same price
- Cloned voices at the same rate
- Signed webhooks and idempotency keys
- Audio hosted for 30 days on a stable URL
- Failed jobs are never charged
- Credit never expires
Top up from $5 to $10,000 once approved. Already a customer? Add credit in Studio.
What that buys
English text, ×1 script weight
- $0.06
A 1,200-word article
~7,000 chars
- $0.46
A 10,000-word long read
~57,000 chars
- $4.12
A 90,000-word audiobook
~515,000 chars
- $56
1,000 articles a month
~7M chars
Script-weighted characters
English speaks about 1,050 characters a minute; Chinese about 300. Weights keep the price per hour of audio level across languages.
- ×1Latin, Cyrillic & Greek scripts$8.00 / M
- ×1.5Arabic, Hebrew & Indic scripts$12.00 / M
- ×2Hangul & Thai$16.00 / M
- ×2.5Kana (Japanese)$20.00 / M
- ×3Han (Chinese)$24.00 / M
Counted with standard Unicode script ranges, so your bill is reproducible from your own request logs. The optional language hint can never raise it, and weights only ever change in your favor.
60 / min
speech jobs created
300 / min
reads, polls & MCP messages
4
concurrent jobs
20,000
characters per request
Safety rails, not credits. Need more headroom? Limits are set per account—email support@everlit.audio.
69 languages, every official EU language included
Word-level timestamps ship today in 25 of them, and coverage keeps expanding. Everything is published live on the languages endpoint.
Latin, Cyrillic & Greek scripts ×1
47 languages- English (word timings available)
- Spanish (word timings available)
- Portuguese (word timings available)
- French (word timings available)
- German (word timings available)
- Italian (word timings available)
- Dutch (word timings available)
- Polish (word timings available)
- Russian (word timings available)
- Ukrainian (word timings available)
- Czech (word timings available)
- Slovak (word timings available)
- Romanian (word timings available)
- Greek (word timings available)
- Hungarian (word timings available)
- Finnish (word timings available)
- Swedish (word timings available)
- Danish (word timings available)
- Bulgarian (word timings available)
- Croatian (word timings available)
- Estonian (word timings available)
- Latvian (word timings available)
- Lithuanian (word timings available)
- Maltese (word timings available)
- Slovenian (word timings available)
- Irish
- Norwegian
- Catalan
- Galician
- Turkish
- Vietnamese
- Indonesian
- Serbian
- Bosnian
- Belarusian
- Kazakh
- Mongolian
- Uzbek
- Georgian
- Icelandic
- Welsh
- Basque
- Esperanto
- Swahili
- Kinyarwanda
- Ganda
- Kabyle
Arabic, Hebrew & Indic scripts ×1.5
17 languages- Arabic
- Hebrew
- Persian
- Urdu
- Uighur
- Kurdish (Sorani)
- Bengali
- Hindi
- Tamil
- Telugu
- Kannada
- Marathi
- Gujarati
- Nepali
- Malayalam
- Punjabi
- Odia
Hangul & Thai ×2
2 languages- Korean
- Thai
Kana (Japanese) ×2.5
1 language- Japanese
Han (Chinese) ×3
2 languages- Chinese
- Cantonese
Your text, verbatim, with a clock on every word
Every finished job can return the words you submitted—punctuation and all—each with a start and end time. They come from your text, not from a speech recognizer’s best guess, so “$5” is one entry whose span covers the spoken “five dollars.”
- Karaoke-style read-along highlighting
- Captions and subtitles that match the script exactly
- Seek to any sentence in a long recording
- Included with every job, at no extra cost
{
"object": "speech.timings",
"id": "spch_01jx...",
"duration_seconds": 58.3,
"words": [
{ "word": "Hello", "start": 0.00, "end": 0.31 },
{ "word": "from", "start": 0.38, "end": 0.55 },
{ "word": "Everlit.", "start": 0.61, "end": 1.12 }
]
}The Everlit Voice AI model, on its own
The same model, pacing and quality checks that narrate for newsrooms and publishers every day, with none of the publishing attached. Send text, get finished audio, and everything you’d expect from an API you’re going to put in a nightly pipeline.
Voice cloning by API
POST a 15-second sample with consent on record and the clone joins your catalog. Private to your account, same rate as ours.
Inline markup
Drop a pause, play a music bed or duck one under a passage, and hand a quote to a second voice, all inside the text you send.
Formats and pacing
MP3, WAV or Opus. Speed from 0.5× to 2× and a paragraph-pause length, with word timings scaled to match.
Captions
SRT, WebVTT or JSON cues built from the job's timings, in one request. They line up with the audio, speed included.
Estimates first
Price a request and predict its duration before you create anything. Same validation as a real job, nothing queued or billed.
Pronunciation library
Per-request overrides for a one-off, or account-wide rules that apply to every future job and article.
Shareable links
Every finished job carries a signed, key-free link that plays in a browser for 24 hours. Hand it to a person, not a script.
OpenAI-compatible
A synchronous /v1/audio/speech endpoint. Point the OpenAI SDK or any tool with a base-URL setting at Everlit and keep your code.
69 languages
One catalog spanning Latin, Cyrillic, Greek, Arabic, Hebrew, Indic, Hangul, Thai, kana, and Han scripts. Mixed-script text just works.
Signed webhooks
Pass a callback URL and we POST the finished job, HMAC-signed, with retries for 30 minutes. No polling loop required.
Stable audio URLs
The download URL is known the instant you submit and stays valid for 30 days. Store it once; no expiring links to refresh.
Idempotency keys
Retry a create safely. The same key and body returns the original job with no second charge, for 30 days.
Hash-only API keys
Keys are shown once and stored as a SHA-256 digest. Revoke instantly from Studio; every response carries rate-limit headers.
Agent-ready
Every speech feature is also an MCP tool, and the OpenAPI 3 document plus Markdown docs are ready for code generators and assistants.
Asynchronous on purpose
Jobs finish in about 15 to 60 seconds because every file gets a full render and an automatic quality check. That makes Everlit ideal for articles, archives, audiobooks, courses, and publish-time pipelines—and the wrong choice for live conversational agents or IVR that need speech in milliseconds. If your app can wait half a minute, you get richer audio for less.
Three steps to your first MP3
Already an Everlit Studio customer? Connect the MCP server with OAuth right now; speech is included on your plan. For the TTS API, demand is high, so we review every request to keep capacity where our customers expect it. Once you’re approved everything else happens in Studio.
Request access
Tell us what you're building and roughly how much text you'll send. With current demand we review every request by hand, usually within a business day.
Add credit, mint a key
Once approved, open Billing → TTS API in Studio, add a card and starting credit from $5, then create an api_ key under API → Keys.
Send text
POST to /v1/speech, or connect the MCP server and ask your agent to publish, narrate or curate. Usage shows up in Studio within minutes.
Ship audio this afternoon.
One URL for your agent, one endpoint for your code. Connect the MCP server and your first published article is a sentence away; take a key and your first narration is a curl away.
What you’ll get
- A 52-tool MCP server: articles, playlists, podcast feeds, pronunciations, voices, music, usage
- OAuth connect for people, static keys for CI and headless agents
- Studio-quality narration as MP3, WAV or Opus, powered by the Everlit Voice AI model
- 69 languages, cloned voices, word timings and captions in 25
- Webhooks, idempotency, OpenAI-compatible endpoint, 30-day stable URLs
- Speech included on paid Studio plans; otherwise $8 per million characters from $5