FlowSpeech for OpenClaw

openclaw-flowspeech is an external speech-provider plugin that connects OpenClaw to FlowSpeech, a context-aware text-to-speech service with voice selection and style prompting.

The plugin follows OpenClaw's public SpeechProviderPlugin contract. It keeps the API key in OpenClaw configuration or the FLOWSPEECH_API_KEY environment variable and sends it only to the official https://flowspeech.io host.

Features

• Registers the flowspeech speech provider with OpenClaw
• Supports 30 FlowSpeech voices, with Kore as the default
• Maps [[tts:voice=...]] directives to FlowSpeech voice selection
• Maps allowed prompt, style, or emotion voice settings to the FlowSpeech prompt
• Converts audio/L16 PCM responses into standard WAV files
• Rejects redirects, non-HTTPS URLs, lookalike hosts, oversized audio, and malformed responses
• Does not bundle API keys or other credentials

Install

After the ClawHub release is approved:

openclaw plugins install clawhub:openclaw-flowspeech

For local package testing:

npm install
npm pack
openclaw plugins install npm-pack:./openclaw-flowspeech-0.1.0.tgz --force

Configure

Set the API key in the process environment:

export FLOWSPEECH_API_KEY="your_api_key"

Then select the provider in OpenClaw TTS configuration:

{
  "messages": {
    "tts": {
      "provider": "flowspeech",
      "providers": {
        "flowspeech": {
          "voiceName": "Kore"
        }
      }
    }
  }
}

An API key can also be supplied through OpenClaw's secret-aware provider configuration. Do not commit it to this repository.

Usage

Use OpenClaw's standard TTS command or automatic TTS mode. When model overrides are enabled, a message may select a voice:

[[tts:voice=Puck]] Welcome to the product update.

When voice settings are allowed, a style prompt is also supported:

[[tts:style=calm and reassuring]] Your report is ready.

Provider configuration

Development

Use Node.js 22.22.3+, 24.15+, or 25.9+.

npm install
npm run check
npm test
npm run build
npm pack --dry-run

Tests use mocked HTTP responses and contain no live provider credentials.

Security

• The plugin never logs the API key.
• baseUrl must be HTTPS and must resolve to the exact flowspeech.io hostname.
• Redirect following is disabled so authorization headers cannot be forwarded to another host.
• Response sizes are bounded before audio is decoded.

Report security issues privately to support@flowspeech.io rather than opening a public issue containing credentials.

License

MIT