Doubao Open TTS SKILL

Overview

This SKILL provides text-to-speech (TTS) capabilities for AI Agents using the Doubao (Volcano Engine) API. It enables agents to convert text into natural-sounding speech with 200+ voice options.

Skill Information

• Name: doubao-open-tts
• Type: Text-to-Speech Synthesis
• Provider: Volcano Engine (Doubao)
• Version: 1.0.0
• Developer: xdrshjr

Capabilities

Core Functions

1. Text-to-Speech Synthesis

   • Convert any text to natural speech
   • Support for multiple audio formats (mp3, pcm, wav)
   • Adjustable speech speed and volume

2. Voice Selection

   • 200+ voices across multiple categories
   • Interactive voice selection workflow
   • Support for Chinese and multilingual voices

3. Voice Categories

   • General (Normal & Multilingual with emotions)
   • Roleplay (characters, personalities)
   • Video Dubbing (cartoon characters)
   • Audiobook (storytelling voices)
   • Customer Service (professional voices)
   • Fun Accents (regional dialects)

Agent Integration

# Example: Agent using this SKILL
from skills.doubao_open_tts import VolcanoTTS, get_voice_selection_prompt, find_voice_by_name

# Step 1: Show voice options to user
prompt = get_voice_selection_prompt()
# Agent displays: "Please select a voice..."

# Step 2: User selects voice (e.g., "Shiny")
user_choice = "Shiny"
voice_type, voice_name = find_voice_by_name(user_choice)

# Step 3: Agent synthesizes speech
tts = VolcanoTTS(app_id="...", access_token="...", secret_key="...")
audio_path = tts.synthesize(
    text="Hello, I'm your AI assistant",
    voice_type=voice_type,
    output_file="response.mp3"
)

Configuration

Required Parameters

Optional Parameters

Setup

1. Install Dependencies

   pip install -r requirements.txt
   

2. Configure Credentials

   cp .env.example .env
   # Edit .env with your Volcano Engine credentials
   

3. Get API Credentials

   • Visit Volcano Engine Console
   • Enable "Doubao Voice" service
   • Create application to get AppID, Access Token, and Secret Key

Usage Examples

Basic Synthesis

from skills.doubao_open_tts import VolcanoTTS

tts = VolcanoTTS()
audio = tts.synthesize("Hello world", output_file="output.mp3")

Interactive Voice Selection

from skills.doubao_open_tts import get_voice_selection_prompt, find_voice_by_name

# Get prompt for user
prompt = get_voice_selection_prompt()
print(prompt)

# Parse user selection
voice_type, name = find_voice_by_name("Shiny")

With Custom Parameters

audio = tts.synthesize(
    text="Custom speech",
    voice_type="zh_male_sunwukong_mars_bigtts",  # Monkey King voice
    speed=1.2,
    volume=0.8,
    encoding="wav"
)

Voice Categories

Popular Voices

• 灿灿/Shiny (Default) - General purpose, Chinese/English
• 猴哥 - Monkey King character voice
• 快乐小东 - Cheerful male voice
• 霸道总裁 - Dominant CEO character

Category Overview

• General-Multilingual: 20+ voices with emotion support
• Roleplay: 20+ character voices
• Video Dubbing: Cartoon and character voices
• Audiobook: Storytelling optimized voices
• Customer Service: Professional service voices
• Fun Accents: Regional dialects (Cantonese, Sichuan, etc.)

File Structure

doubao-open-tts/
├── README.md              # This file
├── doubao-open-tts.md     # Detailed documentation
├── requirements.txt       # Python dependencies
├── .env.example          # Configuration template
└── scripts/
    ├── tts.py            # Main SKILL implementation
    └── test_tts.py       # Test and example scripts

Agent Workflow Integration

Typical Use Case

1. User Request: "Read this article aloud"
2. Agent Action:
   • Call get_voice_selection_prompt()
   • Present voice options to user
3. User Response: "Use the cheerful female voice"
4. Agent Action:
   • Call find_voice_by_name("cheerful female")
   • Get voice_type
   • Call tts.synthesize(article_text, voice_type=voice_type)
5. Result: Return audio file to user

Dependencies

• volcengine-python-sdk - Volcano Engine SDK
• python-dotenv - Environment variable management

License

MIT License

Support

• Issues: GitHub Issues
• Developer: @xdrshjr
• Documentation: See doubao-open-tts.md for complete API reference