KdaiP

EchoBot. EchoBot: Your Anime AI Companion

StableTTS. Next-generation TTS model using flow-matching and DiT, inspired by Stable Diffusion 3

yolov8-deepsort-tracking. opencv+yolov8+deepsort行人检测与跟踪,以及可选的WebUI界面（基于gradio）

labelme2YoloV8-segment. 将labelme数据标注格式转换为YoloV8语义分割数据集，并可自动划分训练集和验证集

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

MobileSAM-fast-finetuning. Finetune MobileSAM with Less Than 4GB RAM!

AutoEncoder-pytorch. 逐行解释的pytorch自编码器实现，使用MNIST数据集进行训练，保证代码简单。

conformer-RoPE. Conformer block with Rotary Position Embedding, modified from lucidrains' implement

1D-Condition-method-pytorch. Conditioning and feature fusion methods such as FiLM, Conditional Layer Norm and AdaIN.

DC-Speech-VAE2-Inference. A 48 kHz speech VAE with 10 Hz frame rate.

contentvec-no-fairseq. Standalone contentvec inference code without the need of installing fairseq

JiOu-LLM. JiOu-LLM: 基于llama2的奇偶数判别模型

Qwen3-ASR-Latents-for-VoiceConversion. Wrapper of Qwen3-ASR for extracting semantic audio features for voice conversion tasks.

ASRT_microphone. 基于ASRT的实时麦克风语音识别项目