Read audio with specific sampling rate and option for mixdown

[frankenjoe]

Some libraries offer the option to set a target rate when reading an audio file, see e.g. librosa.load(). I wonder if we should also give the user the option to request a specific sampling rate and maybe also the option for a mixdown using our audresample package. This could be quite convenient for users not aware of aware of audresample .

[hagenw]

I'm not really in favor of adding a dependency to audresample to audiofile. This was also the reason why I closed #108.

At the moment I see two issues with audresample:

• the wheel that is downloaded is much bigger than it needs to be until Build different Python packages for different architectures audresample#25 is solved
• we cannot easily support all available platforms with audresample as we have seen with OSError when importing opensmile on MacBook M1 laptop opensmile-python#64

As an alternative solution I would propose that we extend the usage documentation, and maybe even the docstring examples of audiofile.read() with examples how to use audresample to resample and/or mix the returned signal.

[frankenjoe]

Yes, those are indeed two strong arguments for avoiding the dependency, though users who don't care much about dependencies may prefer librosa.load() if they want to resample. But I guess that's ok. At least extending the docstring seems like a good idea.