Tencent TTS Podcast Generator

Convert text content to podcast audio files using Tencent Cloud TTS service.

Capabilities

What This Skill Can Do

• Short & Long Text Compatible: Intelligently detects text length, processes short text directly, auto-chunks long text
• Long Text to Speech: Supports generating podcasts up to 30 minutes long (~7200 characters)
• Concurrent Processing: Long texts are automatically split and processed in parallel for faster generation
• 26 Voices: Supports basic, featured, customer service, and Tencent featured voices
• Smart Chunking: Splits text at semantic boundaries (paragraph/sentence) for natural audio flow
• Duration Estimation: Automatically estimates generated audio duration
• Auto Retry: Automatically retries failed requests to improve success rate

Short & Long Text Processing Strategy

> Note: Tencent Cloud TTS single request limit is ~150 characters. Texts exceeding this will be auto-chunked.

What This Skill Does NOT Do

• Does not generate mp3 format (wav only)
• Does not support background music or sound effects
• Does not auto-generate podcast scripts (user must provide)
• Does not support dual-speaker dialogue mode (single voice only)

File Structure

This Skill consists of the following files:

• tts_podcast.py

Main entry script

• Tencent Cloud TTS signature generation
• Audio file generation
• COS upload functionality

• tts_tool.py

AgentScope tool interface wrapper

• SKILL.md

This file, describing Skill capabilities, boundaries, and usage conventions

• requirements.txt

Python dependency configuration

Input & Output Specifications

Input Parameters

Output

{
  "Code": 0,
  "Msg": "success",
  "AudioUrl": "https://xxx.cos.ap-guangzhou.myqcloud.com/xxx.wav"
}

Usage

Environment Requirements

• Python 3.8+
• tencentcloud-sdk-python
• cos-python-sdk-v5
• requests

Install Dependencies

pip install -r requirements.txt

Basic Usage

from tts_podcast import main

result = main({
    "Text": "Hello, welcome to today's podcast.",
    "VoiceType": 502006,
    "secret_id": "YOUR_SECRET_ID",
    "secret_key": "YOUR_SECRET_KEY"
})

print(result)
# {'Code': 0, 'Msg': 'success', 'AudioUrl': 'https://...'}

Short Text Optimized Usage

# Short text (<150 chars) - Use single thread for fast response
result = main({
    "Text": "Hello, this is a short message.",
    "VoiceType": 502006,
    "secret_id": "YOUR_SECRET_ID",
    "secret_key": "YOUR_SECRET_KEY",
    "max_workers": 1,      # Single thread is sufficient
    "timeout": 30,         # 30 second timeout
    "enable_retry": True   # Enable retry
})

Long Text Optimized Usage

# Long text (>150 chars) - Use concurrency for speed
long_text = """Chapter 1: The Origin of AI

The concept of artificial intelligence can be traced back to ancient Greek mythology..."""

result = main({
    "Text": long_text,
    "VoiceType": 502007,
    "secret_id": "YOUR_SECRET_ID",
    "secret_key": "YOUR_SECRET_KEY",
    "max_workers": 5,          # Concurrent processing
    "chunk_size": 140,         # 140 chars per chunk
    "timeout": 60,             # 60 second timeout
    "preserve_paragraphs": True  # Preserve paragraph boundaries
})

Voice Reference

Technical Architecture

tts_podcast.py

• TTS: Uses Tencent Cloud TTS API signature v3
• Upload: Uses Tencent Cloud COS SDK for audio file upload
• Auth: Supports credentials from parameters or environment variables
• Short & Long Text Compatible:
• Short text (≤150 chars): Direct single request, fast response
• Long text (>150 chars): Smart chunking + concurrent processing + auto-merge

Text Chunking Strategy

1. Paragraph Priority: Try to preserve paragraph integrity, split at paragraph boundaries
2. Sentence Boundaries: When paragraphs are too long, split at sentence ending punctuation (。！？；)
3. Semantic Protection: Avoid truncating in the middle of words, ensure semantic coherence
4. Length Control: Each chunk does not exceed 150 characters (Tencent Cloud API limit)

License

MIT