ElevenLabs

Add voice synthesis to FlowRunner workflows with ElevenLabs. Agents convert text to natural speech, generate sound effects, clone and design voices, and store the audio for downstream use.

17 actions available

A new row lands in a content sheet with a script column

Agent reads the script text and the target voice from the row

Agent confirms the requested voice exists in the account

Text converted to speech and returned as an audio file URL

URL written back to the source row for the content pipeline

Production channel notified that the audio is ready

Branded voice-overs routed to a reviewer before they are published

What This Integration Enables

Agents generate speech from text, produce sound effects from a description, and transcribe audio into text, with every output stored and returned as a URL. They manage the voice library directly: instant and professional cloning, designing a voice from a text description, editing voices, and adding samples. Helper actions return the available voices and models as dictionaries, so a flow can validate or select a voice and model before generating. History actions retrieve and clean up past generations. The connector covers audio creation and voice management; the surrounding flow decides where each file goes.

Without FlowRunner

Audio is a separate tool Each voice-over is produced by hand in a standalone app and exported

Transcripts lag Recorded audio waits for someone to run it through a transcription step

Inconsistent brand voice Voice selection is left to whoever produces each asset

With FlowRunner

Audio produced in the flow Speech and sound effects are generated inline the moment the flow needs them

Transcription on arrival Audio is transcribed as part of the workflow, not as a later chore

One defined voice, reused A designed brand voice is referenced by ID across the whole content library

Use Case Scenarios

Audio Content Pipeline

A content team maintains a sheet where each row holds a script and a target voice. A Google Sheets new-row trigger fires. The agent confirms the voice exists with Get Voices, calls Text to Speech, and writes the returned audio URL back into the row. It posts a note to the production channel that the file is ready. The team reviews finished audio instead of producing each clip by hand.

Brand Voice Prototype to Production

A marketing operations lead wants a consistent brand voice across a content library. The agent calls Design Voice from Text to prototype candidates, saves the chosen one with Create Voice from Generation, and references that voice ID in every subsequent Text to Speech call. One designed voice drives the whole library instead of a different sound on each asset.

Meeting Audio to Emailed Transcript

An audio recording arrives in storage. The agent calls Speech to Text to transcribe it, formats the transcript, and passes it to Brevo's Send Transactional Email so stakeholders receive the text. A branded voice-over in the same flow is held for review: the agent generates the audio but routes it to the brand owner before it publishes.

Human-in-Loop Highlight

A cloned or branded voice carries reputational weight, so the agent treats published voice-overs as a human decision. When an agent generates audio in the company's branded voice for external publishing, it pauses and routes the file to the brand owner via Slack: "Generated voice-over for the launch announcement using the brand voice. Listen and approve for publish, regenerate, or reject?" The owner hears exactly what will go out and decides. Internal transcripts and routine audio flow through automatically; anything that speaks for the brand gets a human in the loop first.

Agent processes routinely

Detects exception requiring judgment

Clear match Continues automatically

Ambiguous Routes to human via Slack

Human decides

Agent resumes with decision

Agent Capabilities

17 actions

Audio Generation

Text to Speech Converts text into natural-sounding speech and returns an audio file URL. Used to produce voice-overs, narration, and voice alerts inline in a flow.
Text to Sound Effects Generates a sound effect from a text description. Used in media production pipelines that need custom audio cues.
Speech to Text Transcribes an audio file into text. Used to turn recordings into searchable, routable transcripts.

Voice Management

Create Voice (Instant Clone) Creates a cloned voice quickly from a small set of samples. Used to stand up a usable voice fast for prototyping.
Create Voice (Professional) Creates a high-fidelity professional cloned voice. Used when production-quality cloning is required.
Create Voice from Generation Saves a generated voice design as a reusable voice. Used to promote a prototyped voice into the library.
Design Voice from Text Prototypes a brand voice from a text description. Used to explore candidate voices before committing to one.
Add Voice Samples Adds audio samples to an existing voice. Used to refine a clone with more reference material.
Edit Voice Updates the settings or metadata of an existing voice. Used to tune or relabel a voice in the library.
Delete Voice Removes a voice from the account. Used to retire a voice that is no longer in use.
Get Voice Retrieves a single voice by ID. Used to confirm a voice's details before generating.
Get Voices Lists the voices available to the account. Used to enumerate options for selection.

Models and History

Get Models Lists the available speech models. Used to choose a model for a generation.
Get User Info Retrieves the authenticated account details. Used to confirm the connected account.
Get History Retrieves the list of past audio generations. Used to audit or reuse prior output.
Get History Item Audio Retrieves the audio file for a specific history item. Used to pull a previously generated file back into a flow.
Delete History Item Removes a generated audio item from history. Used to clean up after a generation is no longer needed.

Start building with ElevenLabs

$100 in credits. No card required. Connect in minutes.

Start Building Free Talk to us