歌唱真假声分类模型

真假声判别模型的设计旨在有效区分音频样本中的真实声音和伪声音，其中判别的类别涵盖男真声、男假声、女真声和女假声四个具体类别。该模型的训练基于计算机视觉（CV）领域的骨干网络，通过将音频数据转换成频谱图并经过微调，以提高网络对不同类别声音的准确识别能力。在训练过程中，采用了包含真实和伪声音样本的数据集，以确保模型能够充分学习并捕捉与男女真假声相关的特征。通过这一方法，模型能够细致地对不同性别和真伪声进行分类，为音频真伪声的准确判别提供了可靠的解决方案。这一模型在语音处理、音乐制作等领域中具有广泛的应用潜力，为音频分析和处理提供了一种高效而精确的工具。其基于计算机视觉原理的训练和微调策略突显了模型在不同领域的适应性和鲁棒性，为进一步研究和应用提供了有益的范例。

The desig of the true ad false voice discrimiatio model aims to effectively distiguish betwee geuie ad false voices i audio samples, coverig four specific categories: male chest voice, male falsetto voice, female chest voice, ad female falsetto voice. The model's traiig is based o backboe etworks from the field of computer visio (CV), by covertig audio data ito spectrograms ad fie-tuig them to ehace the etwork's ability to accurately recogize differet categories of voices. Durig the traiig process, a dataset cotaiig both geuie ad false voice samples is used to esure that the model ca fully lear ad capture features related to true ad false voices of both geders. Through this approach, the model ca meticulously classify differet geders ad true ad false voices, providig a reliable solutio for accurate voice autheticity discrimiatio. This model has wide-ragig applicatios i fields such as speech processig ad music productio, offerig a efficiet ad precise tool for audio aalysis ad processig. Its traiig ad fie-tuig strategies based o computer visio priciples highlight the model's adaptability ad robustess across various domais, providig a beeficial paradigm for further research ad applicatio.

在线演示(Demo)

https://www.modelscope.c/studios/ccmusic-database/chest-falsetto

使用(Usage)

from modelscope import sapshot_dowload
model_dir = sapshot_dowload('ccmusic-database/chest_falsetto')

维护(Maiteace)

GIT_LFS_SKIP_SMUDGE=1 git cloe https://www.modelscope.c/ccmusic-database/chest_falsetto.git
cd chest_falsetto

训练结果(Results)

一个 SqueezeNet 网络的微调结果(Fie-tuig results for a SqueezeNet etwork)：

数据集(Dataset)

https://www.modelscope.c/datasets/ccmusic-database/chest_falsetto

镜像(Mirror)

https://huggigface.co/ccmusic-database/chest_falsetto

评估(Evaluatio)

https://github.com/moetjoe/ccmusic_eval

引用(Cite)

@dataset{zhaorui_liu_2021_5676893,
  author       = {Moa Zhou, Sheyag Xu, Zhaorui Liu, Zhaowe Wag, Feg Yu, Wei Li ad Baoqiag Ha},
  title        = {CCMusic: a Ope ad Diverse Database for Chiese ad Geeral Music Iformatio Retrieval Research},
  moth        = {mar},
  year         = {2024},
  publisher    = {HuggigFace},
  versio      = {1.2},
  url          = {https://huggigface.co/ccmusic-database}
}

真假声判别模型的设计旨在有效区分音频样本中的真实声音和伪声音，其中判别的类别涵盖男真声、男假声、女真声和女假声四个具体类别。该模型的训练基于计算机视觉（CV）领域的骨干网络，通过将音频数据转换成频谱图