***的开源音频（波形等）工具，你要找的是什么jMIR的JAUDIO子集：的 http://jmir.sourceforge.net/jAudio.html 。你可以使用的项目，虽然它不是开放源代码是免费使用的***的工具，是EchoNest API。我假设当你说音乐你的意思是音响;如果它的乐谱，你可以，如果数据是MIDI还是我music21工具包。特点模块，如果它是另一种格式使用jMIR的jSymbolic工具。

Is there any open-source code can extract content-based music features like tempo, beats, etc.?

My application scenario is like this:

1. A music repository which contains more than 20,000 songs

2. Extract features and auto-tag these songs by the "Program" which could be suggested by you.

3. Auto classify these songs by the tags.

The best open-source audio (waveform, etc.) tool for what you're looking for is the jAudio subset of jMIR: http://jmir.sourceforge.net/jAudio.html. The best tool that you could use for the project, though it's not open-source is free to use, is the EchoNest API. I am assuming that when you say "music" you mean audio; if it's musical scores, you could use the jSymbolic toolkit of jMIR if the data is MIDI or the .features module of my music21 toolkit if it's another format.