erhu_playing_tech

简介 Intro

二胡演奏技法识别模型是一种基于深度学习技术的音频分析工具, 旨在自动区分二胡演奏中的不同技法。该模型通过深入分析二胡音乐的声学特征, 能够识别出包括分弓、垫弓、泛音、连弓、滑音、大滑音、击弓、拨弦、抛弓、顿弓、颤弓、颤音以及揉弦在内的11种基本演奏技法。通过对音频信号进行时频域转换、特征提取和模式识别, 该模型能够准确地对二胡演奏中的复杂技法进行分类, 为音乐信息检索、音乐教育以及二胡演奏艺术的研究提供了一种高效的技术支持。此模型的应用不仅丰富了音乐声学领域的研究, 也为传统音乐的传承与创新开辟了新的途径。

The Erhu Performance Technique Recognition Model is an audio analysis tool based on deep learning techniques, aiming to automatically distinguish different techniques in erhu performance. By deeply analyzing the acoustic characteristics of erhu music, the model is able to recognize 11 basic playing techniques, including split bow, pad bow, overtone, continuous bow, glissando, big glissando, strike bow, pizzicato, throw bow, staccato bow, vibrato, tremolo and vibrato. Through time-frequency conversion, feature extraction and pattern recognition, the model can accurately categorize the complex techniques of erhu performance, which provides an efficient technical support for music information retrieval, music education, and research on the art of erhu performance. The application of this model not only enriches the research in the field of music acoustics, but also opens up a new way for the inheritance and innovation of traditional music.

在线演示 Demo (推理代码)

使用 Usage

:modelscope-code[]{type="sdk"}

维护 Maintenance

:modelscope-code[]{type="git"}

训练结果 Results

Backbone Mel CQT Chroma

:----------------: :----------------------------------: :---------: :---------:

Swin-S 0.978 0.940 0.903

Swin-T ** 0.994 ** ** 0.958 ** ** 0.957 **

AlexNet 0.960 0.970 0.933

ConvNeXt-T ** 0.994 ** ** 0.993 ** ** 0.954 **

ShuffleNet-V2-X2.0 0.990 0.923 0.887

GoogleNet 0.986 0.981 0.908

SqueezeNet1.1 0.932 0.939 0.875

Average 0.976 0.958 0.917

最佳结果 Best Result

一个 Swin-T 网络在 mel 谱上的微调结果(Fine-tuning results for a Swin-T network on mel)：

Loss curve

Training and validation accuracy

Confusion matrix

数据集 Dataset

镜像 Mirror

校验 Evaluation

引用 Cite