🐸💬 - a deep learning toolkit for Text-to-Speech, battle-tested in research and production

🤖 💬 Deep learning for Text to Speech (Discussion forum: https://discourse.mozilla.org/c/tts)

Speaker embedding (d-vector) trained with GE2E loss

zero-shot realtime TTS system, fully offline, free and open source

[InterSpeech 2020] "Improving the Speaker Identity of Non-Parallel Many-to-Many VoiceConversion with Adversarial Speaker Recognition" by Shaojin Ding, Guanlong Zhao, Ricardo Gutierrez-Osuna

Real-Time-Voice-Cloning using Generative Adversarial Networks. It combines text and speaker and generates natural sounding audio. Some Traditional Methods like Convolutional Neural Networks can contain some limitation it requires large datasets and lengthy processing time, to overcome these limitations we are using GANs to maximize space use.

🎤 Create high-quality text-to-speech audio using FastSpeech 2 in PyTorch. Easily customize for English and Mandarin with multi-speaker support.

Create fast, lightweight text-to-speech audio on your CPU with MioTTS-llama.cpp. Convert text to WAV files using customizable voice options.

Generate multilingual speech and clone voices with this 0.6B-parameter zero-shot text-to-speech model.

Build low-latency voice agents using a modular pipeline with an OpenAI-compatible WebSocket API.

Run Qwen3-TTS locally using Intel Arc GPUs and IPEX for fast, hardware-accelerated voice cloning.

Run pure C inference for Mistral Voxtral-4B-TTS with mmap safetensors, BF16 weights, and 24kHz WAV output.

Generate zero-shot multilingual TTS nodes for ComfyUI with voice cloning and voice design, supporting 600+ languages

Build text-to-speech applications with this curated guide for real-time agent streaming and high-fidelity offline synthesis.