🐍📦 Ultra-fast Python package for calculating and analyzing the Word Error Rate (WER). Built for the scalable evaluation of speech and transcription accuracy.

Ming-UniAudio: Speech LLM for Joint Understanding, Generation and Editing with Unified Representation

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

轻量级大语言模型MiniMind的源码解读，包含tokenizer、RoPE、MoE、KV Cache、pretraining、SFT、LoRA、DPO等完整流程

A python module to repair invalid JSON from LLMs

Open-Source Frontier Voice AI

Simultaneous speech-to-text model

MOSS-TTSD is a spoken dialogue generation model that enables expressive dialogue speech synthesis in both Chinese and English, supporting zero-shot multi-speaker voice cloning, and long-form speech…

Text-audio foundation model from Boson AI

Chinese voice corpus. 中文语音语料，语音更加清晰自然，包含8个开源数据集，3200个说话人，900小时语音，1300万字。

UTokyo-SaruLab MOS Prediction System

A Collection of Papers on Diffusion Language Models

Spark-TTS Inference Code

Voice Activity Detector (VAD) : low-latency, high-performance and lightweight

中文文本纠错数据集汇总

A Benchmark for Evaluating Turn-Taking and Overlap Handling in Full-Duplex Spoken Dialogue Models

Kimi-Audio, an open-source audio foundation model excelling in audio understanding, generation, and conversation

Fine-tune the Whisper speech recognition model to support training without timestamp data, training with timestamp data, and training without speech data. Accelerate inference and support Web deplo…

📰 Must-read papers and blogs on LLM based Long Context Modeling 🔥

Paper list of simultaneous translation / streaming translation, including text-to-text machine translation and speech-to-text translation.

[EMNLP 2025] MultiMed-ST: Large-scale Many-to-many Multilingual Medical Speech Translation

A large-scale speech corpus introduced in Spark-TTS, built from diverse open-source datasets for training text-to-speech (TTS) systems.

代码大模型 预训练&微调&DPO 数据处理 业界处理pipeline sota

Heuristic filtering framework for RefineCode

Towards Human-Sounding Speech

A quick guide (especially) for trending instruction finetuning datasets