slng

n8n community node for SLNG: text-to-speech, speech-to-text, and a trigger that registers a webhook tool on a SLNG voice agent

Package Information

Released: 1 mo ago
Downloads: 0 weekly / 0 monthly
Latest Version: 0.4.1
Author: Nicolas Grenié

Available Nodes

Documentation generating…

SLNG icon
SLNG
Use SLNG voice AI to generate text-to-speech audio or transcribe speech-to-text from audio files
SLNG Trigger icon
SLNG Trigger
Start workflows from SLNG voice agents by registering a secure webhook tool for agent calls

Documentation

slng.ai

n8n-nodes-slng

This is an n8n community node. It lets you use SLNG in your n8n workflows.

SLNG is a unified voice AI platform offering text-to-speech, speech-to-text, and voice agents that can place and receive phone calls.

n8n is a fair-code licensed workflow automation platform.

Installation
Operations
Credentials
Compatibility
Usage
Templates
Resources
Version history

Installation

In your n8n instance, go to Settings → Community Nodes → Install, enter the package name n8n-nodes-slng, and confirm. See the installation guide for details.

For a self-hosted instance you can also install it manually:

mkdir -p ~/.n8n/nodes
cd ~/.n8n/nodes
npm install n8n-nodes-slng

Restart n8n after installing or upgrading.

Operations

This package contains two nodes:

SLNG

  • Text to Speech → Generate — convert text into spoken audio. The audio is returned as a binary file on the item.
  • Speech to Text → Transcribe — transcribe an audio file (from a binary field) into text.
  • Agent → Dispatch Call — place an outbound phone call from a SLNG voice agent. The node checks that the selected agent has outbound telephony configured before dispatching.

The Model and (for TTS) Voice fields are dropdowns populated live from the SLNG catalog API (GET /v1/catalog/models); the voice list is filtered to the chosen TTS model. Switch any of them to By ID to hardcode a model path or voice ID.

SLNG Trigger

Registers a webhook tool on one or more SLNG agents when the workflow is activated, and removes it on deactivation. When an agent calls the tool during a call, the workflow runs and (in Using Last Node mode) returns its output to the agent.

On activation the node creates an org-level SLNG tool, publishes it, and attaches the published version to every agent you select in the Agents field. Two things to know:

  • Agents must be in shared tool mode. The picker lists all your agents; legacy-mode agents are marked "(legacy, unsupported)" and will error on activation, since they cannot hold shared tool attachments.

  • Activation runs the workflow once with sample data. Publishing an API-request tool requires SLNG's publish "green run", which sends one test call to the webhook URL. Your workflow therefore executes a single time with placeholder arguments when you activate it.

  • Tool type — LLM tool (contextual): the agent decides when to call it; you define the parameters it sends with a simple field builder (name, type, description, required). System tool: fires automatically on a call lifecycle event (call start, first user message, call end, tool succeeded/failed) with worker-populated arguments.

  • Show Advanced Settings — a toggle. Off by default with secure defaults (HMAC auth + auto-generated secret, POST, return last node's output). Turn it on to expose webhook path, authentication, HTTP method, response handling, result instructions and more.

  • Webhook authentication — HMAC (SLNG signs the body with X-Signature-256) or Bearer (Authorization: Bearer). Leave the secret empty to auto-generate one on activation; the node registers it with SLNG and validates incoming requests.

  • Respond (advanced) — Using Last Node (default) returns the workflow's final node output to the agent; Immediately acknowledges and runs the workflow in the background. To shape the exact JSON the agent receives, keep Using Last Node and make the final node (e.g. a Set / Edit Fields node) emit the object you want. (n8n's built-in Respond to Webhook node only works with core trigger types, not community triggers, so it isn't supported here.)

If SLNG sends an envelope payload with an arguments field, the trigger exposes it to the workflow as toolArguments. n8n blocks expression access to $json.arguments for security reasons, so downstream nodes should use {{$json.toolArguments}} or {{$json.toolArguments.fieldName}}.

To reuse data from a previous test while mapping downstream nodes, use n8n's Pin Data feature on the SLNG Trigger output. A good setup flow is: run one real SLNG test call, pin the trigger output data, then build the rest of the workflow against toolArguments.

Credentials

You need a SLNG API key. Create one in the SLNG dashboard, then add a SLNG API credential in n8n and paste the key. The key is sent as a Bearer token and is validated against GET https://api.slng.ai/v1/me. The same credential is used for both the Voice API (api.slng.ai) and the Agents API (api.agents.slng.ai).

Compatibility

Built against the n8n nodes API version 1. Requires Node.js 18+.

Usage

  1. Add the SLNG API credential.
  2. To expose a workflow to a voice agent, add a SLNG Trigger node, pick an existing agent, define the tool, and activate the workflow. The tool is added to the agent automatically.
  3. To synthesize, transcribe, or dispatch outbound agent calls inside any workflow, use the SLNG node.

Templates

Multi-node workflow templates live in templates/. They demonstrate practical SLNG voice-agent automations:

  • End-of-call logging to Notion and Slack.
  • Mid-call Linkup web search.
  • Mid-call HubSpot customer lookup.
  • Post-call CRM note, Gmail follow-up, and Slack notification.
  • Outbound call campaign from HubSpot contacts.

Import them into n8n with Workflows → Import from File, then replace placeholder IDs and assign your credentials.

Resources

Version history

See CHANGELOG.md for the full release history.

Contributing

Development and release instructions live in CONTRIBUTING.md.

Discussion