Parler TTS

Parler-TTS is a powerful, open-source text-to-speech (TTS) library designed for generating high-quality, natural-sounding speech. It allows users to control speech features like gender, pitch, and style through simple text prompts. Parler-TTS is fully open-source, providing access to datasets, training code, and model weights, enabling community-driven development.

The library supports two new checkpoints, an 880M and a 2.3B parameter model, trained on 45k hours of audiobook data. With optimizations like SDPA and Flash Attention 2, Parler-TTS offers faster generation times.

Installation is straightforward with a single command, and the library is compatible with Apple Silicon. Users can experiment with different speaker characteristics from a list of 34 pre-trained voices.

Key Features:

• Customizable Speech: Control voice features using text prompts.
• Open-Source: Access to all resources for community development.
• Optimized Performance: Faster generation with advanced optimizations.

Whether you're building applications or conducting research, Parler-TTS offers a robust platform for high-quality TTS model development.