Tortoise Audio Book Creator - Create audiobooks 100% locally for FREE - Text to speech (TTS)

Introduction

Creating high-quality audiobooks has been a quest for many, including myself, George. Since 2002, I have explored various text-to-speech (TTS) engines, investing in numerous services without achieving satisfactory results until recently. The introduction of neural network voices marked a turning point in the quality of TTS, with platforms like Azure and Google Cloud leading the way, but costs remained a concern.

I sought a local solution to avoid ongoing expenses and finally discovered Tortoise Text-to-Speech, a local solution that met my long-standing needs.

Discovering Tortoise TTS

The evolution of AI-driven TTS models has opened many doors. After trying multiple options such as Bark and Coqui, I found Tortoise Text-to-Speech to be a robust contender. Alongside it, I explored a faster version known as Tortoise Fast, which significantly enhanced performance. However, the original Tortoise model had its shortcomings, including artifacts in synthesis and control issues with voice selection.

With effective hardware, I utilized powerful NVIDIA graphics cards (RTX 390 and RTX 490) to run Tortoise and embarked on creating audiobooks—something I'd dreamed of for over two decades.

Installation Steps

Here's how you can set up your Tortoise Audio Book Creator on Ubuntu Linux (22.04):

1. Install Ubuntu and Update: After installing Ubuntu, update your system with the following commands:

   sudo apt update
   sudo apt upgrade
   

2. Install Proprietary Drivers: Navigate to 'Additional Drivers' and install the latest drivers for your graphics card.

3. Install Necessary Packages: Ensure you have the required packages installed.

4. Set Up Conda Environment:

   • Download the Conda installer for Linux.
   • Make it executable using:

     chmod +x <installer_filename>
     

   • Run the installer (without sudo).

5. Clone the Repository:

   • Navigate to your desired directory and clone the repository:

     git clone <repository_url>
     

6. Create Conda Environment: In the cloned repository folder, run:

   conda env create -f environment.yml
   

7. Activate Environment: Activate the Conda environment with:

   conda activate <env_name>
   

8. Download Necessary Models: Obtain the auto-regressive model files and place them in the specified cache directory.

9. Run Tortoise: Launch the Tortoise application and adjust settings as needed to generate audiobooks.

Key Features

• Self-Correction Rounds: This feature enhances audio quality by using Whisper to identify discrepancies in generated audio compared to the original text.
• Voice Latent Files: Creating voice latent files from hours of recording helps improve the quality of generated speech, making it sound more natural.

Text Formatting for Better Output

Properly formatting your text before processing it through Tortoise is crucial. Use pauses (in milliseconds) to enhance the listening experience—this helps the spoken text flow more naturally.

Conclusion

Utilizing Tortoise Audio Book Creator enables anyone to generate audiobooks locally and for free. With the right setup and a bit of experimentation, you can produce high-quality audiobooks that rival commercial options.

Now, let's break down this article further.

---

Keywords

• Tortoise Audio Book Creator
• Text-to-Speech
• Neural Network Voices
• Audiobook Production
• Local TTS Solutions
• Voice Latent Files
• Self-Correction Rounds

---

FAQ

1. What is Tortoise Audio Book Creator?

• Tortoise Audio Book Creator is a text-to-speech tool that allows users to create audiobooks locally and for free, leveraging advanced AI models.

2. What operating system is needed for installation?

• The installation has been tested on Ubuntu Linux 22.04.

3. What hardware specifications are recommended?

• Powerful NVIDIA graphics cards, such as RTX 390 or RTX 490, are recommended for optimal performance.

4. How do I improve the audio quality of my audiobooks?

• Utilizing voice latent files and self-correction rounds when processing can significantly enhance audio quality.

5. Can I run Tortoise without an internet connection?

• Yes, Tortoise runs locally, which means it does not require continuous internet access once set up properly.