For Agents
Generate speech in cloned or library voices, build custom voices, transcribe audio, detect deepfakes and watermarks, and dispatch Resemble voice agents to phone numbers.
Use for: Read this paragraph aloud in our brand voice, Check whether this audio clip is a deepfake, Transcribe the uploaded meeting recording and summarise the action items, Create a new voice from a description of a calm older narrator
Not supported: Does not host podcasts, edit video, or translate text between languages. Use for Resemble speech synthesis, voice creation, transcription, deepfake and watermark detection, and voice agents only.
The Resemble AI API synthesizes speech from text or SSML in cloned and library voices, synchronously or as an HTTP stream, and builds custom voices from uploaded recordings or from a text prompt through voice design. It also transcribes audio and answers questions about a transcript, edits and enhances audio, and runs deepfake detection on a media url with optional audio source tracing and follow-up questions. Further operations apply and detect audio watermarks, search speaker identities, run phone-dispatched voice agents with tools, webhooks, and knowledge items, and manage pronunciations, term substitutions, projects, and clips.
Install Jentic One Beta
Jentic One is a self-hosted execution layer for AI agents. It lets your agent call the API Reference, or any other public or private API you need. You set the rules, the agent never sees your credentials, and every call is logged.
Two steps, two machines. Install the instance in a safe environment, then register your agent from wherever it runs.
Step 1: Jentic One Host machine
# On the machine that will host your Jentic One instance:
curl -fsSL "https://jentic.com/install.sh?src=apis&api=%2Fapis%2Fresemble.ai%2Fresemble" | shStep 2: Agent machine
# On the machine where your agent runs (keep this separate from the instance):
curl -fsSL "https://jentic.com/install.sh?src=apis&api=%2Fapis%2Fresemble.ai%2Fresemble" | sh
jentic register # connects your agent to your Jentic One instanceJentic One is in public beta. The setup above keeps your agent separate from the instance, which is what you want before using real credentials: an agent running as the same OS user as Jentic One can read its stored keys directly. Just evaluating? A single local install is fine to start. See the secure deployment guide for the tiers.
What an agent can do with API Reference API.
Synthesize up to 3,000 characters of text or SSML in a chosen voice_uuid, with an optional chatterbox-turbo model for lower latency
Stream synthesized speech over HTTP for real-time playback
Build a custom voice from uploaded recordings or generate voice candidates from a text user_prompt
Run deepfake detection on an audio or video url, then ask follow-up questions about the result
Apply an audio watermark to a file and detect watermarks in suspect media
Transcribe audio and ask intelligence questions about a transcript
Dispatch a Resemble voice agent to a phone_number with dynamic_variables, tools, and knowledge items
Patterns agents use API Reference API for, with concrete tasks.
★ Agent voiceover generation
An agent connected through Jentic turns a script into audio by calling the synchronous synthesis operation with a voice_uuid and the text or SSML in data, choosing output_format and sample_rate. Longer narration is split into chunks under the 3,000-character limit, and custom pronunciations can be applied with apply_custom_pronunciations.
Call POST /synthesize with voice_uuid set to the brand voice and data set to the first 3,000-character chunk of the script
Deepfake screening
Trust and safety teams submit a media url to the detection operation, optionally with audio_source_tracing or a callback_url, then read the result and ask questions about it through the detection intelligence operations. Watermark detection adds a second check for content produced with Resemble watermarking.
Call POST /detect with url set to the suspect clip, then GET /detect/{uuid} until the result is ready
Phone voice agents
Teams create a Resemble agent with a voice, attach tools, webhooks, and knowledge items, and dispatch it to a phone_number with dynamic_variables for the call. Agent phone numbers are managed through their own operations.
Call POST /agents/{uuid}/dispatch with phone_number set to +15551234567 and dynamic_variables carrying the customer's first name
Custom voice creation
A product team creates a voice by name, uploads recordings to it, and triggers a build, or generates candidates from a user_prompt with voice design and promotes one to a voice.
Call POST /voice-design with user_prompt set to 'calm older male narrator', then POST /voice-design/create with the chosen candidate
109 endpoints — the resemble ai api synthesizes speech from text or ssml in cloned and library voices, synchronously or as an http stream, and builds custom voices from uploaded recordings or from a text prompt through voice design.
METHOD
PATH
DESCRIPTION
/synthesize
Synchronous text-to-speech synthesis
/stream
Streaming text-to-speech over HTTP
/voices
Create a voice
/voice-design
Generate voice candidates from a prompt
/detect
Create a deepfake detection
/watermark/apply
Apply an audio watermark
/speech-to-text
Create a transcript job
/agents/{uuid}/dispatch
Dispatch a voice agent to a phone number
/synthesize
Synchronous text-to-speech synthesis
/stream
Streaming text-to-speech over HTTP
/voices
Create a voice
/voice-design
Generate voice candidates from a prompt
/detect
Create a deepfake detection
/watermark/apply
Apply an audio watermark
/speech-to-text
Create a transcript job
/agents/{uuid}/dispatch
Dispatch a voice agent to a phone number
This API is usable in Jentic One now. Its AI-readiness score against Jentic's framework shows where it stands today and where improvements would make it even easier for agents to use.
Base layer of spec validity and structural soundness.
Aggregated quality score from linter diagnostics, weighted by severity.
Percentage of `$ref` references that resolve successfully.
Checks whether the API description parses successfully and conforms to its declared specification (e.g., OpenAPI).
Structural correctness score based on schema issues using logarithmic dampening.
Clarity, completeness, and ingestion readiness for developers and tooling.
How richly the API is illustrated with examples.
Percentage of examples that conform to their schemas.
Percentage of operations with complete response definitions (success, client error, server error).
Health of API ingestion, bundling, and resolution within Jentic pipelines.
Semantic breadth, depth, and agent comprehension for AI systems.
Coverage of descriptions across API elements.
Coverage of RFC 9457 Problem Details for error responses.
Coverage, uniqueness, and casing consistency of operationIds for AI inference.
Coverage of summaries across operations/tags/info.
Functional utility, complexity comfort, and AI orchestration readiness.
Agent comfort level based on API operational and structural complexity.
Trust, risk posture, and security compliance.
Average quality of security schemes based on authentication method strength (weakest link for OAuth2).
Findability, semantic richness, and reasoning readiness.
Clarity and depth of descriptions across API elements.
Score it yourself
Every API in the directory is allowlisted, so you can re-score it with no key required.
npx @jentic/api-scorecard-cli score <openapi-url>What agents get from Jentic-routed access to this vendor.
Setup
Wiring Resemble by hand means sending its bearer token on each call, choosing between the f.cluster.resemble.ai and app.resemble.ai hosts, chunking text under the 3,000-character synthesis limit, and polling detection, transcript, and watermark jobs. Through Jentic you install once, import Resemble AI from the API Directory, store the token once, and your agent calls it.
Permission scoping
Synthesis takes its voice_uuid in the request body, so rules bound which operations your agent may use there, not which voice. Agents, voices, and projects carry their ids in the URL path (/agents/{uuid}/dispatch, /voices/{voice_uuid}), so a rule can tie the agent to one of them, and destructive ones like voice or project deletion are not included unless you add them.
Credential isolation
Your Resemble API token is stored once, encrypted, by your own Jentic One instance and injected at execution time. It never enters the agent's prompt, logs, or context.
Intent-based discovery
Agents search Jentic by intent such as 'read text aloud in a cloned voice' or 'check if audio is a deepfake', and Jentic returns the matching Resemble operation with its input schema so the agent sends the right voice_uuid, url, or phone_number.
Alternatives and complements available in the Jentic catalogue.
Specific to using API Reference API through Jentic.
What authentication does the Resemble AI API use?
Per its OpenAPI spec, the Resemble AI API authenticates with an HTTP bearer API token sent in the Authorization header, and each operation also declares that header as a required parameter. Tokens come from the API page of your Resemble account. Through Jentic, the token is stored encrypted by your own Jentic One instance and injected at execution time.
How much text can I synthesize in one Resemble call?
POST /synthesize takes up to 3,000 characters of text or SSML in data, plus a required voice_uuid. Optional fields include model, output_format, sample_rate, precision, use_hd, and project_uuid. For real-time playback use POST /stream instead.
Can I detect deepfakes with the Resemble AI API?
Yes. POST /detect takes a media url and options such as audio_source_tracing, zero_retention_mode, and callback_url, and GET /detect/{uuid} returns the result. You can then ask questions about a detection with POST /detects/{uuid}/intelligence.
What are the rate limits for the Resemble AI API?
The OpenAPI spec does not publish rate limit numbers. Check https://docs.resemble.ai for current limits.
How do I generate speech through Jentic?
Search Jentic for 'synthesize speech with Resemble'. Jentic returns POST /synthesize with its input schema; your agent sets voice_uuid and data, and Jentic sends the call with your stored token. To run it on your own infrastructure, install Jentic One from its GitHub repo.
Is there a Resemble AI MCP server?
You don't need an MCP server to give your agent Resemble AI. Jentic connects it directly from the API Directory: import it, store your token once, and your agent finds the synthesis, detection, and agent operations by intent instead of loading all of their tool definitions into context.
Can I limit what my agent is allowed to do with the Resemble AI API?
Yes. A narration agent could be allowed POST /synthesize, POST /stream, and GET /voices only, so it can speak in existing voices but cannot create or delete voices, dispatch phone agents, or remove projects. For a calling workflow, allow POST /agents/{uuid}/dispatch for one agent uuid.
GET STARTED