└── README.md /README.md: -------------------------------------------------------------------------------- 1 | # Speech-Simulation-Tools 2 | 语音增强领域的相关数据仿真工具和方法汇总--持续更新 3 | 4 | 1：MASG, https://github.com/vipchengrui/MASG 5 | 功能实现：房间声学中的麦克风阵列语音发生器(MASG)，仿真室内环境下的麦克风阵列采集数据 6 | 7 | 2：gpuRIR, https://github.com/DavidDiazGuerra/gpuRIR 8 | 功能实现：gpu加速的RIR仿真 9 | 10 | FAST-RIR: Fast neural diffuse room impulse response generator 11 | FAST-RIR：快速神经扩散房间脉冲响应发生器 12 | 链接：https://arxiv.org/abs/2110.04057 13 | 作者：Anton Ratnarajah,Shi-Xiong Zhang,Meng Yu,Zhenyu Tang,Dinesh Manocha,Dong Yu 14 | 机构： University of Maryland, College Park, MD, USA, Tencent AI Lab, Bellevue, WA, USA 15 | 备注：https://anton-jeran.github.io/FRIR/ https://github.com/anton-jeran/FAST-RIR 16 | 17 | 3：StoRIR，https://github.com/SRPOL-AUI/storir 18 | 功能实现：一种不同于RIR的随机房间脉冲响应生成方法，用于机器学习应用中的音频数据增强，不需要知道房间尺寸和声源位置 19 | 20 | 4：RIR，https://github.com/ehabets/RIR-Generator 21 | 功能描述：基于matlab版本的RIR生成算法仿真 22 | 23 | 5: DNS-Challenge,https://github.com/microsoft/DNS-Challenge 24 | 功能实现：微软开源的DNS挑战赛的数据和仿真脚本，可以直接进行数据的仿真，包括加混响和噪声 25 | 26 | 6：ConferenceSpeech2021,https://github.com/ConferencingSpeech/ConferencingSpeech2021/tree/master/eval 27 | 功能实现：腾讯组织的麦克风阵列语音增强比赛，其中的eval可以批处理生成pesq,stoi，si-sdr等客观评价指标，并输出文件，非常有用。 28 | 注：非常好的多通道仿真数据生成脚本，可以根据干净语音、噪声和冲击响应生成响应的多通道数据，包括混响参考数据和非混响参考数据，以及对应的带噪语音数据。 29 | 1：仿真训练所需数据时，给出干净语音，多通道冲击响应和噪声后，即可仿真生成多通道带噪混响语音； 30 | 2：训练时，默认为8通道数据，每个通道6s，配置时train.yaml中val_chun和train_chunk均为6 31 | pyrirgen库备用安装工程：https://github.com/YoungJay0612/rir-generator-pyrirgen 32 | 33 | 7: AECMOS: A speech quality assessment metric for echo impairment 34 | 标题：AECMOS：一种回声损伤的语音质量评估指标 35 | 链接：https://arxiv.org/abs/2110.03010 36 | 作者：Marju Purin,Sten Sootla,Mateja Sponza,Ando Saabas,Ross Cutler 37 | 机构：Microsoft Corporation 38 | 39 | # Speech Enhancement Measurement 40 | link：https://github.com/IMLHF/Speech-Enhancement-Measures 41 | speech enhancement metrics：CSIG, CBAK, CMOS, SSNR, PESQ, STOI, ESTOI, SNR, IS, LLR, WSS 42 | 43 | link:https://github.com/kamo-naoyuki/pySIIB 44 | pySIIB: A python implementation of speech intelligibility in bits (SIIB) 45 | 46 | # Audio Generation 47 | Audio generation using diffusion models, in PyTorch.基于diffusion模型的音频生成 48 | link: https://github.com/archinetai/audio-diffusion-pytorch#experiments 49 | 50 | # Feature Extraction 特征提取 51 | link: https://github.com/YoungJay0612/padasip 52 | 通过Python进行Adaptive Signal Processing 53 | 54 | link: https://github.com/bootphon/shennong 55 | Shennong: A Python toolbox for speech features extraction 56 | 57 | link：https://github.com/ZhihaoDU/speech_feature_extractor 58 | GFCC等特征提取 59 | 60 | # Loss Compute 61 | Link: https://github.com/csteinmetz1/auraloss 62 | auraloss: A collection of audio-focused loss functions in PyTorch. 63 | 64 | # Data Augmentation 数据增强 65 | 66 | ### 2021-8-9 67 | paper title: SpecMix : A Mixed Sample Data Augmentation method for Training withTime-Frequency Domain Features 68 | paper title: SpecMix：一种利用时频域特征训练的混合样本数据增强方法 69 | 链接：https://arxiv.org/abs/2108.03020 70 | 作者：Gwantae Kim,David K. Han,Hanseok Ko 71 | github： https://github.com/GT-KIM/specmix 72 | 机构：Korea University, South Korea, Drexel University, USA 73 | 备注：Accepted to interspeech 2021 74 | 75 | muda: https://github.com/bmcfee/muda https://muda.readthedocs.io/en/latest/ 76 | SpecAugment: https://github.com/DemisEom/SpecAugment 77 | 78 | Facebook-Augly: https://github.com/facebookresearch/AugLy 79 | Introc: AugLy is a data augmentations library that currently supports four modalities (audio, image, text & video) and over 100 augmentations. 80 | 81 | Asteroid-team: https://github.com/asteroid-team/torch-audiomentations 82 | Introc: Audio data augmentation in PyTorch. Inspired by audiomentations. 83 | 84 | Audiomentation: https://github.com/iver56/audiomentations 85 | Introc: A Python library for audio data augmentation. Inspired by albumentations（图像数据增强）. 86 | --------------------------------------------------------------------------------