node-red-contrib-say

Node-RED output node for text-to-speech (TTS).
Let your runtime speak messages from your flows.

What this node does

node-red-contrib-say adds a say node to the output section of the Node-RED palette.

When the node receives a message it:

• picks the text to speak from:
  • Text field in the node config, or
  • node Name, or
  • msg.payload
• chooses a voice (built-in or custom string, depending on your platform)
• chooses a speed factor
• uses the underlying say library to speak the text on the host machine
• forwards the original msg to the next node when speaking finishes

This makes it easy to:

• announce sensor values (“Temperature is 21 degrees”)
• speak notifications on a kiosk / wallboard
• give audible feedback from flows on your local machine or home server

Compatibility

| | Version | | --- | --- | | Supported Node-RED | 4.x (recommended; aligned with current Node-RED releases) | | Node.js | 18.x or newer (CI tests 20 and 22 on Ubuntu and macOS) |

Older Node-RED or Node.js versions may still work but are not covered by automated tests.

Requirements

This node is a thin wrapper around the say package and relies on platform TTS tools:

• macOS

  • Uses the built-in say command.
  • No extra setup required in a default macOS install.
  • To list voices available on that machine, run in a terminal: say -v '?' (voices can differ by macOS version and installed language packs).

• Linux

  • The underlying say package uses Festival only (for example espeak-ng is not used).

  • Install Festival using your package manager (for example on Debian/Ubuntu):

    sudo apt-get install festival

  • You also need a working audio stack on the host (for example ALSA, PulseAudio, or PipeWire) so Festival can play sound. Minimal or headless servers may need extra configuration or audio devices mapped into the environment.

  • Docker and containers: Base images often omit Festival and may not expose sound devices. Install Festival inside the image, install/configure audio, and pass through devices or use a suitable remote audio setup—otherwise the node cannot produce audible output.

• Windows

  • Uses a native PowerShell call to SAPI.SpVoice.
  • Works out‑of‑the‑box on standard Windows installations with PowerShell.

The Node-RED container / host must have at least one working TTS backend installed and accessible from the command line, otherwise this node cannot speak.

Installation

You can install the node either from the Node-RED palette or from your project directory.

From the Node-RED palette

1. Open the Node-RED editor.
2. Go to Menu → Manage palette → Install.
3. Search for node-red-contrib-say.
4. Click Install.

From your project directory

In your Node-RED user directory (often ~/.node-red or your project folder):

pnpm add node-red-contrib-say

Then restart Node-RED so the new node is loaded.

Example flows

This package ships importable flows under examples/ (see Node-RED: packaging examples):

| File | What it shows | | --- | --- | | examples/01-say-fixed-text.json | Inject → say → debug — text comes from the node Text field. | | examples/02-say-from-payload.json | Inject → say → debug — Text empty; spoken string is msg.payload. |

To import: Menu → Import → select a JSON file (or paste the file contents), then Deploy. Ensure your runtime has a working TTS setup (see Requirements).

Using the say node

After installation you will find the say node under Output in the Node-RED palette.

Inputs

• msg.payload
  • Used as the text to speak if neither the Text field nor the node Name is set.
  • Can be a string, number, or anything that can be converted to a string.

Outputs

• 1 output
  • The original msg is forwarded once speaking is finished or an error occurred.
  • On error the node will also log via node.error(err).

Node configuration

The editor dialog (say.html) exposes the following fields:

• Name

  • Optional label shown under the node in the flow.
  • If Text is empty, the Name is also used as the spoken text (fallback).

• Text

  • Optional text to speak.
  • Highest priority:
    • If set, this value is spoken regardless of msg.payload.
  • Leave empty if you want to drive the spoken text entirely via msg.payload.

• Voice

  • Dropdown of available built‑in voices for your platform. The presets refer to the Node-RED runtime OS (Linux vs macOS), not necessarily the computer where you edit the flow.
  • The editor automatically hides non-matching preset groups when the runtime OS can be detected.
  • If runtime OS detection is unavailable, all preset groups remain visible as a safe fallback.
  • If a flow already uses a preset from another OS, that selected preset stays visible in the editor for compatibility.
  • Options:
    • Standard (empty value): use the default system voice.
    • Specify voice by string (:): enables Voice Name.
    • Named voices for Linux (Festival voices like Alan, Nick, SLT, …).
    • Named voices for macOS (e.g. Alex, Bruce, Kathy, Vicki, …).

• Language

  • Filters preset voices by locale (for example de_DE, en_US, fr_FR) on top of runtime OS filtering.
  • Default can be All languages.
  • The editor remembers the last selected language filter locally and restores it the next time you open the dialog.
  • Standard and Specify voice by string remain available regardless of active language filter.

• Voice Name (shown only when Voice is set to Specify voice by string)

  • Free‑form string passed to the underlying TTS engine as the voice name.
  • Use this when your platform has additional voices that are not listed in the dropdown.

• Speed

  • Playback speed factor.
  • Default: 1.0.
  • Range in the UI: 0.1 – 2.0, step 0.1.
  • Internally converted to a Number and passed to say.speak.

Text selection priority

When a message arrives, the node chooses the text to speak in this order:

1. config.text (Text field), if not empty
2. config.name (Name), if Text is empty
3. msg.payload, if neither Text nor Name are provided

Example flows

1. Speak a fixed message on deploy or button press

1. Add an Inject node.
2. Set it to send any payload (e.g. "trigger").
3. Add a say node.
4. Open the say node dialog and set:
   • Text: Hello from Node-RED
   • Voice: Standard
   • Speed: 1.0
5. Wire Inject → say and deploy.

Whenever you click the inject button, your machine will say:
“Hello from Node-RED”.

2. Speak dynamic payload values

1. Add an Inject node and configure:
   • Payload type: string
   • Payload: Temperature is 21 degrees
2. Add a say node with:
   • leave Text empty
   • optional Name: Speak temperature
3. Wire Inject → say and deploy.

On each inject, the say node will speak the string from msg.payload.

3. Use a custom voice string

1. On a platform that supports multiple voices (macOS / Festival), add a say node.
2. In the node configuration:
   • Set Voice to Specify voice by string.
   • Set Voice Name to a valid voice identifier, e.g. Alex or another voice name recognised by your OS.
3. Leave Text empty and provide text via an Inject or Function node.

The node will now try to speak using the custom voice you configured.

Error handling

• If the underlying say call fails, the node:
  • calls node.error(err) for visibility in the debug sidebar / logs,
  • does not send the message further (no node.send) in that case.
• Make sure the host has a working TTS engine installed and available on the command line.

Contributing

Development setup, linting, and git hooks are described in CONTRIBUTING.md.

License & credits

• License: MIT – see LICENSE.
• Based on: say by Marak – huge thanks for the original library.
• Built for: Node-RED, an open-source flow-based programming tool for wiring together hardware devices, APIs and online services.