tts-providers

A config-driven text-to-speech tool for OpenClaw. One script, multiple TTS providers, zero dependencies.

Adding a new provider = adding a JSON block. No code changes.

Features

• 🔧 Config-driven — providers defined in tts_providers.json
• 🗣️ Multi-provider — DashScope (Qwen3-TTS), Speechify, easily extensible
• 📦 Zero dependencies — pure Python stdlib
• 🔑 Reads API keys from environment (OpenClaw env.vars compatible)
• 🌏 Multilingual — English, Chinese, Japanese, and more

Quick Start

# Set your API key
export DASHSCOPE_API_KEY="sk-..."

# Default provider
echo "Hello world" | python3 tts.py -o hello.mp3

# Specify provider
python3 tts.py -p dashscope --text "Hello" -o hello.mp3
python3 tts.py -p speechify --text "Hello" --voice george -o hello.mp3

# Instruction-controlled expressiveness
python3 tts.py -p dashscope-instruct --instruct "Speak warmly" --text "Welcome" -o warm.mp3

# List providers
python3 tts.py --list-providers

# List voices
python3 tts.py --list-voices -p dashscope

Adding a New Provider

Edit tts_providers.json and add a block:

{
  "my-provider": {
    "name": "My TTS Provider",
    "endpoint": "https://api.example.com/v1/tts",
    "auth": {"header": "Authorization", "prefix": "Bearer", "env": "MY_API_KEY"},
    "defaults": {"model": "default-model", "voice": "default-voice"},
    "request_template": {
      "model": "{{model}}",
      "text": "{{text}}",
      "voice": "{{voice}}"
    },
    "response": {
      "type": "url",
      "path": "output.audio.url"
    }
  }
}

Response types

Type	Description
url	Response contains an audio URL at the configured path. Script downloads it.
base64	Response contains base64-encoded audio at the configured path. Script decodes it.

Template variables

Variable	Source
{{text}}	Input text
{{voice}}	--voice or provider default
{{model}}	--model or provider default
{{lang}}	--lang or provider default
{{instruct}}	--instruct (stripped if empty)

Bundled Providers

Provider	Model	Key	Notes
dashscope	Qwen3-TTS Flash	DASHSCOPE_API_KEY	International endpoint (Singapore)
dashscope-instruct	Qwen3-TTS Instruct Flash	DASHSCOPE_API_KEY	Instruction-controlled expressiveness
speechify	Simba English	SPEECHIFY_API_KEY	Requires paid subscription

OpenClaw Integration

Add API keys to openclaw.json → env.vars:

{
  "env": {
    "vars": {
      "DASHSCOPE_API_KEY": "sk-...",
      "SPEECHIFY_API_KEY": "..."
    }
  }
}

License

MIT