🌐🤖 SpeechStylis AI: The Cutting-Edge in Text-to-Speech Synthesis with Python! 🚀🔊

Hey there, tech enthusiasts! 👋 Are you tired of boring, robotic text-to-speech voices? 😴 Well, we've got some exciting news for you! 🎉 We're introducing SpeechStylis AI, the cutting-edge technology that's revolutionizing the world of text-to-speech synthesis with Python! 🚀

Imagine being able to generate natural-sounding speech from text input, with a tone and style that matches your personality or brand. 💬 That's exactly what SpeechStylis AI does! It uses advanced machine learning algorithms to analyze a large dataset of human speech recordings, and then generates new speech samples that sound like they were recorded by a real person. 🤯

Ready to give it a try? SpeechStylis AI is now available as a Python library, so you can easily integrate it into your own projects. 🛠️ Whether you're building a virtual assistant, creating an audiobook, or developing an accessibility tool, SpeechStylis AI has everything you need to make your vision a reality. 🏡

💬 Where to ask questions

Please use our dedicated channels for questions and discussion. Help is much more valuable if it's shared publicly so that more people can benefit from it.

Type	Platforms
🐛 Bug Reports	GitHub Issue Tracker
🎁 Feature Requests & Ideas	GitHub Issue Tracker
💻 Usage Questions	GitHub Discussions
🗨️ General Discussion	GitHub Discussions or Discord

🚀 Features

• Pretrained Models: Explore a wide range of pretrained models in over 1100 languages.

• Versatile Tools: Utilize tools for training new models and fine-tuning existing ones in any language.

• Dataset Analysis: Leverage utilities for dataset analysis and curation.

Model Implementations

Spectrogram models

• Tacotron: paper
• Tacotron2: paper
• Glow-TTS: paper
• Speedy-Speech: paper
• Align-TTS: paper
• FastPitch: paper

Installation

👩‍💻SpeechStylis AI is tested on Ubuntu 18.04 with python >= 3.9, < 3.12.

Tested Platforms:

• Ubuntu
• Kali Linux
• Google Cloud

pip install TTS

If you are on Ubuntu (Debian) or Kali Linux, you can also run following commands for installation.

git clone https://github.com/haydenbanz/SpeechStylis.git

Modify .py File to Specify Pre-recorded Audio Location

To use your prerecorded audio, locate the .py file and find the section where the speaker's WAV file path is defined. Update the speaker_wav_path variable with the path to your audio file. Below is an example:

# Original Code
speaker_wav_path = "/content/drive/MyDrive/audio.wav"

Run in Google Colab

You can run SpeechStylis AI in Google Colab, which provides a free and convenient environment for running the notebook. Click the badge below to open the SpeechStylis.ipynb notebook in Colab:

Best Practices for Voice Cloning

To achieve the best results with voice cloning, it's important to use high-quality audio samples. The quality of the input audio directly affects the quality of the generated speech. Here are some tips for creating effective audio samples:

• Use a High-Quality Microphone: A good microphone will capture your voice with clarity and detail, which is crucial for the model to learn the nuances of your voice. Avoid using built-in laptop microphones if possible.

• Record in a Quiet Environment: Background noise can interfere with the voice cloning process. Record in a quiet room with minimal echo or reverberation.

• Speak Naturally: Speak in a natural and consistent tone. Avoid whispering or shouting. The model will learn from your speaking style, so be sure to provide a good example.

• Provide Sufficient Audio: The more audio you provide, the better the model will be at cloning your voice. Aim for at least 30 seconds of clear, continuous speech.

• Use a Variety of Sentences: Reading a variety of sentences will help the model learn the different sounds and intonations of your voice.

Synthesizing speech by SpeechStylis AI

Choosing a Model

SpeechStylis AI supports a variety of text-to-speech models, each with its own strengths. Choosing the right model depends on your specific needs for quality, speed, and language support.

Here are some recommended models:

• High-Quality Multilingual Voice Cloning (Recommended for Quality):

  • Model: tts_models/multilingual/multi-dataset/your_tts
  • Description: This model is excellent for cloning voices in different languages. It's a bit slower but offers outstanding results.
  • Use Case: Ideal for applications where voice quality and naturalness are top priorities.

• Faster English-Only Model:

  • Model: tts_models/en/ljspeech/tacotron2-DDC
  • Description: If you only need English and want faster generation, this model is a good choice.
  • Use Case: Suitable for applications where speed is more important than multilingual support.

• Fairseq Models for a Wide Range of Languages:

  • Model: tts_models/<lang-iso_code>/fairseq/vits
  • Description: These models support over 1100 languages but may be slower. Replace <lang-iso_code> with the desired language code (e.g., "deu" for German).
  • Use Case: Perfect for projects that require broad language support.

Example Usage

To use a specific model, you can modify the SpeechStylis.py file. Here's an example of how to use the your_tts model for high-quality voice cloning:

import torch
from TTS.api import TTS

# Get device
device = "cuda" if torch.cuda.is_available() else "cpu"

# Initialize TTS with the desired model
tts = TTS(model_name="tts_models/multilingual/multi-dataset/your_tts", progress_bar=True).to(device)

# Text to be converted to speech
text = "Hello, world! This is a test of the new and improved SpeechStylis AI."

# Path to a high-quality audio file of the speaker's voice
speaker_wav_path = "my/cloning/audio.wav"

# Language of the text
language = "en"

# Generate speech and save it to a file
output_file_path = "output.wav"
tts.tts_to_file(text=text, speaker_wav=speaker_wav_path, language=language, file_path=output_file_path)

print(f"Text-to-speech completed. Audio saved to: {output_file_path}")

Running a single speaker model

# Init TTS with the target model name
tts = TTS(model_name="tts_models/de/thorsten/tacotron2-DDC", progress_bar=False).to(device)

# Run TTS
tts.tts_to_file(text="Ich bin eine Testnachricht.", file_path=OUTPUT_PATH)

# Example voice cloning with YourTTS in English, French and Portuguese
tts = TTS(model_name="tts_models/multilingual/multi-dataset/your_tts", progress_bar=False).to(device)
tts.tts_to_file("This is voice cloning.", speaker_wav="my/cloning/audio.wav", language="en", file_path="output.wav")
tts.tts_to_file("C'est le clonage de la voix.", speaker_wav="my/cloning/audio.wav", language="fr-fr", file_path="output.wav")
tts.tts_to_file("Isso é clonagem de voz.", speaker_wav="my/cloning/audio.wav", language="pt-br", file_path="output.wav")

Example voice conversion

Converting the voice in source_wav to the voice of target_wav

tts = TTS(model_name="voice_conversion_models/multilingual/vctk/freevc24", progress_bar=False).to("cuda")
tts.voice_conversion_to_file(source_wav="my/source.wav", target_wav="my/target.wav", file_path="output.wav")

Example voice cloning together with the voice conversion model.

This way, you can clone voices by using any model

tts = TTS("tts_models/de/thorsten/tacotron2-DDC")
tts.tts_with_vc_to_file(
    "Wie sage ich auf Italienisch, dass ich dich liebe?",
    speaker_wav="target/speaker.wav",
    file_path="output.wav"
)

Example text to speech using Fairseq models in ~1100 languages 🤯.

For Fairseq models, use the following name format: tts_models/<lang-iso_code>/fairseq/vits. You can find the language ISO codes here and learn about the Fairseq models here.

# TTS with on the fly voice conversion
api = TTS("tts_models/deu/fairseq/vits")
api.tts_with_vc_to_file(
    "Wie sage ich auf Italienisch, dass ich dich liebe?",
    speaker_wav="target/speaker.wav",
    file_path="output.wav"
)

Command-line tts

Synthesize speech on command line.

You can either use your trained model or choose a model from the provided list.

If you don't specify any models, then it uses LJSpeech based English model.

If you have any questions or feedback, please contact the project maintainers:

• 0x_hayden
• Email: t5hlt8zcp@mozmail.com

Credits

This project is maintained by:

Contributors and Developers

Support

If you find this project helpful, consider buying us a coffee:

License

SpeechStylis AI is licensed under the Mozilla License. See the LICENSE file for details.