Managed access to microphone input and system volu

问题:

I am looking to do three things:

Access data from the microphone. Really all I want to know is the overall volume of the sound sensed by the device.

Set the microphone gain.

Set the system volume.

All of my windows dev experience is C#/WPF, so I'd like to stay managed. I do not need exceptionally high performance or realtime processing or anything.

I've looked around and it seems like SlimDX might be a good wrapper for this, but even there I'm not sure where to start.

Surely it can't be that hard?

回答1:

Here's a link that shows how to access the audio mixer in Windows from C#:

http://www.codeguru.com/csharp/csharp/cs_graphics/sound/article.php/c10931

This will let you set the microphone gain and the system volume. The first part is a little more complicated, though. Basically, you need to start recording the input (using DirectSound or the waveInXXXX API [my personal favorite]). As each buffer gets filled with audio, you can calculate the Root Mean Square for the buffer and use this to estimate volume.

Edit: here's a link to a project (that I've used and modified successfully, so I know it works) that shows how to record audio using the waveInXXXX API:

http://www.codeproject.com/KB/audio-video/cswavrec.aspx?df=90&fid=16677&mpp=25&noise=3&sort=Position&view=Quick&select=3005817

Edit 2: and since I'm tired of posting links, here's an actual formula for calculating the Root Mean Square of an audio buffer (the type here is float[], but it can be easily modified to handle short[], which is what you'd normally get from waveInXXXX):

public static float RootMeanSquared(ref float[] audio)
{
    double sumOfSquared = 0;
    for (int i = 0; i < audio.Length; i++)
    {
        sumOfSquared += audio[i] * audio[i];
    }
    return (float)Math.Sqrt(sumOfSquared / (double)audio.Length);
}

回答2:

Unfortunately you can't reliably read (or render) data from managed code unless you're willing to live with serious latency (on the order of .5 second). The problem is that the CLR can interrupt your process for 250 milliseconds at a time without warning. Normally this doesn't matter, but when you're trying to do isochronous processing it can be a significant issue.

回答3:

You can use NAudio to capture audio from the microphone in managed C#. Have a look at the demo projects for examples of how to do this. As Larry points out above, don't expect great latency though. NAudio also has managed wrappers for the mixer APIs which should let you set the microphone volume, although it can be tricky to get hold of the correct control programmatically.

回答4:

I've just wrote a (very basic) sample code on how to capture sound from the microphone using SlimDX. I hope it helps if you're still looking for an answer.