diff --git a/README.md b/README.md index 348621e..39808cd 100644 --- a/README.md +++ b/README.md @@ -1,6 +1,7 @@ # 🎙️ 简易播客生成器 (Simple Podcast Generator) > 轻松将您的想法，一键生成为生动有趣的多人对话播客！ +> [English Version](README_EN.md) 这是一个强大的脚本工具，它利用 **OpenAI API** 的智慧生成富有洞察力的播客脚本，并通过 **TTS (Text-to-Speech)** API服务，将冰冷的文字转化为有温度的音频。您只需提供一个主题，剩下的交给它！ @@ -74,10 +75,6 @@ python podcast_generator.py [可选参数] python podcast_generator.py --api-key sk-xxxxxx --model gpt-4o --threads 4 ``` ---- - ---- - ### 4. 自定义 AI 提示词（`custom` 代码块） 为了提供更细致的 AI 指令或添加特定上下文，您可以在 `input.txt` 文件中嵌入 `custom` 代码块。此代码块中的内容将作为额外指示，被内置到播客脚本生成的核心提示词（`prompt-podscript.txt`）之中，从而影响 AI 的生成行为。 @@ -185,16 +182,13 @@ python podcast_generator.py --api-key sk-xxxxxx --model gpt-4o --threads 4 所有成功生成的播客音频文件将自动保存在 `output/` 目录下。文件名格式为 `podcast_` 加上生成时的时间戳，例如 `podcast_1678886400.wav`。 ---- - ---- ## 🎧 示例音频 您可以在 `example/` 文件夹中找到使用不同 TTS 服务生成的播客示例音频： -* **Edge TTS 生成示例**: [edgeTTS_podcast_1754467217.aac](example/edgeTTS_podcast_1754467217.aac) -* **Index TTS 生成示例**: [indexTTS_podcast_1754467749.aac](example/indexTTS_podcast_1754467749.aac) +* **Edge TTS 生成示例**: +* **Index TTS 生成示例**: 这些音频文件展示了本工具在实际应用中的效果。 diff --git a/README_EN.md b/README_EN.md new file mode 100644 index 0000000..51d1c4c --- /dev/null +++ b/README_EN.md @@ -0,0 +1,210 @@ +# 🎙️ Simple Podcast Generator + +> Easily transform your ideas into lively and engaging multi-person conversational podcasts with a single click! +> [中文版](README.md) + +This is a powerful script tool that leverages the intelligence of **OpenAI API** to generate insightful podcast scripts and transforms cold text into warm audio through **TTS (Text-to-Speech)** API services. You just need to provide a topic, and it handles the rest! + +✨ The podcast script generation logic of this project is deeply inspired by the [SurfSense](https://github.com/MODSetter/SurfSense) project. We express our sincere gratitude for its open-source contribution! + +--- + +## ✨ Core Features + +* **🤖 AI-Driven Scripting**: Automatically generate high-quality, in-depth podcast dialogue scripts with the powerful OpenAI model. +* **👥 Multi-Role Support**: Freely define multiple podcast roles (e.g., host, guest) and assign a unique TTS voice to each role. +* **🔌 Flexible TTS Integration**: Seamlessly connect with your self-built or third-party TTS services through simple API URL configuration. +* **🔊 Smart Audio Merging**: Automatically and precisely stitch together voice segments from various roles to synthesize a complete, smooth podcast audio file (`.wav` format). +* **⌨️ Convenient Command-Line Interface**: Provides clear command-line parameters, giving you full control over every aspect of the podcast generation process. + +--- + +## 🛠️ Installation Guide + +### 📝 Prerequisites + +1. **Python 3.x** + * Please ensure Python 3 is installed on your system. + +2. **FFmpeg** + * This project relies on FFmpeg for audio merging. Please visit the [FFmpeg official website](https://ffmpeg.org/download.html) to download and install it. + * **Important**: After installation, please ensure the `ffmpeg` command is added to your system's environment variable (PATH) so that the script can call it normally. + +### 🐍 Python Dependencies + +Open your terminal or command prompt and install the required Python libraries using pip: +```bash +pip install requests openai +``` + +--- + +## 🚀 Quick Start + +### 1. Prepare Input Files + +Before running, please ensure the following files are ready: + +* `input.txt`: Enter the **podcast topic** or core ideas you wish to discuss in this file. +* `prompt/prompt-overview.txt`: A system prompt used to guide AI in generating the **overall outline** of the podcast. +* `prompt/prompt-podscript.txt`: A system prompt used to guide AI in generating the **detailed dialogue script**. It contains dynamic placeholders (e.g., `{{numSpeakers}}`, `{{turnPattern}}`), which the script will automatically replace. + +### 2. Configure TTS Service and Roles + +* The `config/` directory contains your TTS configuration files (e.g., `edge-tts.json`). This file defines the TTS service's API interface, podcast roles (`podUsers`), and their corresponding voices (`voices`). + +### 3. Run the Script + +Execute the following command in the project root directory: + +```bash +python podcast_generator.py [Optional Parameters] +``` + +#### **Optional Parameters** + +* `--api-key `: Your OpenAI API key. If not provided, it will be read from the configuration file or the `OPENAI_API_KEY` environment variable. +* `--base-url `: Proxy address for the OpenAI API. If not provided, it will be read from the configuration file or the `OPENAI_BASE_URL` environment variable. +* `--model `: Specify the OpenAI model to use (e.g., `gpt-4o`, `gpt-4-turbo`). The default value is `gpt-3.5-turbo`. +* `--threads `: Specify the number of parallel threads for audio generation (default is `1`), improving processing speed. + +#### **Run Example** + +```bash +# Use gpt-4o model and 4 threads to generate the podcast +python podcast_generator.py --api-key sk-xxxxxx --model gpt-4o --threads 4 +``` + +### 4. Custom AI Prompts (`custom` code block) + +To provide more detailed AI instructions or add specific context, you can embed `custom` code blocks in the `input.txt` file. The content of this code block will serve as additional instructions, built into the core prompt (`prompt-podscript.txt`) for podcast script generation, thereby influencing the AI's generation behavior. + +**Usage**: +In any location within the `input.txt` file, define your custom content using the following format: + +``` +```custom-begin +Additional instructions or context you wish to provide to the AI, for example: +- "Please ensure the discussion includes an in-depth analysis of [specific concept]." +- "Please add some humorous elements to the dialogue, especially jokes about [a certain topic]." +- "All character speeches must be concise, and each sentence should not exceed two lines." +```custom-end +``` + +**Effect**: +All text content within the `custom` code block (excluding the `custom-begin` and `custom-end` tags themselves) will be extracted and appended to the processed content of the [`prompt/prompt-podscript.txt`](prompt/prompt-podscript.txt) template. This means that these custom instructions will directly influence the AI's decisions and style when generating specific podcast dialogue scripts, helping you to control the output more precisely. + +**Example Scenario**: +If you want the AI to particularly emphasize the future development of a certain technological trend when discussing a tech topic, you can add this to `input.txt`: + +``` +```custom-begin +Please foresightedly analyze the disruptive changes AI might bring in the next five years, and mention the potential impact of quantum computing on existing encryption technologies. +```custom-end +``` + +--- + +## ⚙️ Configuration File Details (`config/*.json`) + +The configuration file is the "brain" of the entire project, telling the script how to work with AI and TTS services. + +```json +{ + "podUsers": [ + { + "code": "zh-CN-XiaoxiaoNeural", + "role": "主持人" + }, + { + "code": "zh-CN-YunxiNeural", + "role": "技术专家" + } + ], + "voices": [ + { + "name": "XiaoMin", + "alias": "晓敏", + "code": "yue-CN-XiaoMinNeural", + "locale": "yue-CN", + "gender": "Female", + "usedname": "晓敏" + }, + { + "name": "YunSong", + "alias": "云松", + "code": "yue-CN-YunSongNeural", + "locale": "yue-CN", + "gender": "Male", + "usedname": "云松" + } + ], + "apiUrl": "http://localhost:5000/api/tts?text={{text}}&voiceCode={{voiceCode}}", + "turnPattern": "random" +} +``` + +* `podUsers`: Defines the **roles** in the podcast. The `code` for each role must correspond to a valid voice in the `voices` list. +* `voices`: Defines all available TTS **voices**. +* `apiUrl`: Your TTS service API endpoint. `{{text}}` will be replaced with the dialogue text, and `{{voiceCode}}` will be replaced with the character's voice code. +* `turnPattern`: Defines the **turn-taking pattern** for character dialogue, such as `random` or `sequential`. + +--- + +## 🔌 TTS (Text-to-Speech) Service Integration + +This project is designed to be highly flexible, supporting various TTS services. Whether locally deployed or cloud-based web services, they can be integrated into this project through simple configuration. + +### 💻 Local TTS Interface Support + +You can deploy the following open-source projects as local TTS services and integrate them into this project via `apiUrl` configuration: + +* **index-tts**: [https://github.com/index-tts/index-tts](https://github.com/index-tts/index-tts) + * **Usage with**: Requires running with `ext/index-tts-api.py`, which provides a simple API interface to encapsulate `index-tts` as a service callable by this project. + +* **edge-tts**: [https://github.com/zuoban/tts](https://github.com/zuoban/tts) + * This is a general TTS library that you can integrate by customizing an adapter. + +### 🌐 Web TTS Interface Support (Pending) + +This project can also be easily configured to integrate various web TTS services. Just ensure your `apiUrl` configuration meets the service provider's requirements. Commonly supported services include: + +* **OpenAI TTS** +* **Azure TTS** +* **Google Cloud Text-to-Speech (Vertex AI)** +* **Minimax TTS** +* **Gemini TTS** (may require integration via custom API adapter) +* **Fish Audio TTS** + +--- + +## 🎉 Output Results + +All successfully generated podcast audio files will be automatically saved in the `output/` directory. The filename format is `podcast_` followed by a timestamp, e.g., `podcast_1678886400.wav`. + +## 🎧 Sample Audio + +You can find sample podcast audio generated using different TTS services in the `example/` folder: + +* **Edge TTS Sample**: +* **Index TTS Sample**: + +These audio files demonstrate the actual effect of this tool in practical applications. + +--- + +## 📂 File Structure + +``` +. +├── config/ # ⚙️ Configuration Files Directory +│ ├── edge-tts.json +│ └── index-tts.json +├── prompt/ # 🧠 AI Prompt Files Directory +│ ├── prompt-overview.txt +│ └── prompt-podscript.txt +├── output/ # 🎉 Output Audio Directory +├── input.txt # 🎙️ Podcast Topic Input File +├── openai_cli.py # OpenAI Command Line Tool +├── podcast_generator.py # 🚀 Main Running Script +└── README.md # 📄 Project Documentation \ No newline at end of file diff --git a/podcast_generator.py b/podcast_generator.py index e3735db..bd2d7df 100644 --- a/podcast_generator.py +++ b/podcast_generator.py @@ -134,8 +134,8 @@ def main(): # Parse command-line arguments parser = argparse.ArgumentParser(description="Generate podcast script and audio using OpenAI and local TTS.") parser.add_argument("--api-key", help="OpenAI API key.") - parser.add_argument("--base-url", help="OpenAI API base URL.") - parser.add_argument("--model", help="OpenAI model to use (e.g., gpt-3.5-turbo).") + parser.add_argument("--base-url", default="https://api.openai.com/v1", help="OpenAI API base URL (default: https://api.openai.com/v1).") + parser.add_argument("--model", default="gpt-3.5-turbo", help="OpenAI model to use (default: gpt-3.5-turbo).") parser.add_argument("--threads", type=int, default=1, help="Number of threads to use for audio generation (default: 1).") args = parser.parse_args()