[ACMMM'2024] Generative Expressive Conversational Speech Synthesis

The most cost-effective, highest performance AI voice agent possible today

MOSS-Speech is a true speech-to-speech large language model without text guidance.

5Hz Deep-Compression Speech VAE for AR-Diffusion and CALMs

A Lightweight and Streaming Zero-Shot Voice Conversion via Mean Flows

A Differentiable Alignment Framework for Sequence-to-Sequence Modeling via Optimal Transport

Lightning-fast, on-device TTS — running natively via ONNX.

Easy, Fast, and Scalable Multimodal AI

The official repo of BridgeVoC, which explores using the Schrödinger Bridge framework for neural vocoding.

A prototype implementation of the "dataset as a queue" pattern for processing web pages into interleaved image/text content.

ProsodyLM: Uncovering the Emerging Prosody Processing Capabilities in Speech Language Models

Text-to-text alignment algorithm for speech recognition error analysis.

High-performance, semantic turn detection for conversational AI

Voice conversion with just linear regression.

A python package for deep multilingual punctuation prediction.

Omnilingual ASR Open-Source Multilingual SpeechRecognition for 1600+ Languages

Emo100Songs: An Open Dataset of Improvised Songs with Emotion Data

TASA & Speech Transformer implementation

This is the official repo for the paper "LongCat-Flash-Omni Technical Report"

This is the repository for the work "BridgeVoC: Revitalizing Neural Vocoder from a Restoration Perspective".

A rigorous framework for evaluating and guiding the development of next-generation AI assistants.

SoulX-Podcast is an inference codebase by the Soul AI team for generating high-fidelity podcasts from text.