/// <summary>
/// Starts capturing audio data.
/// </summary>
/// <param name="targetLatencyInMs">
/// The target maximum number of milliseconds of acceptable lag between
/// live sound being produced and capture operation.
/// </param>
/// <param name="gain">
/// The gain to be applied to the captured audio.
/// </param>
/// <param name="outFormat">
/// The desired output format of the captured audio.
/// </param>
/// <param name="speech">
/// If true, optimizes the audio capture pipeline for speech recognition.
/// </param>
public void StartCapture(int targetLatencyInMs, float gain, WaveFormat outFormat, bool speech)
{
if (this.wasapiCapture != null)
{
this.StopCapture();
}
this.wasapiCapture = new WasapiCapture(this.audioDevice);
// Create a callback delegate and marshal it to a function pointer. Keep a
// reference to the delegate as a class field to prevent it from being GC'd.
this.callbackDelegate = new AudioDataAvailableCallback(this.AudioDataAvailableCallback);
// initialize the capture with the desired parameters
this.wasapiCapture.Initialize(targetLatencyInMs, gain, outFormat, this.callbackDelegate, speech);
// tell WASAPI to start capturing
this.wasapiCapture.Start();
}
/// <summary>
/// Initialize the resampler.
/// </summary>
/// <param name="targetLatencyInMs">
/// The target maximum number of milliseconds of acceptable lag between
/// input and resampled output audio samples.
/// </param>
/// <param name="inFormat">
/// The input format of the audio to be resampled.
/// </param>
/// <param name="outFormat">
/// The output format of the resampled audio.
/// </param>
/// <param name="callback">
/// Callback delegate which will receive the resampled data.
/// </param>
public void Initialize(int targetLatencyInMs, WaveFormat inFormat, WaveFormat outFormat, AudioDataAvailableCallback callback)
{
// Buffer sizes are calculated from the target latency.
this.bufferLengthInMs = targetLatencyInMs;
this.inputBytesPerSecond = (int)inFormat.AvgBytesPerSec;
this.inputBufferSize = (int)(this.bufferLengthInMs * inFormat.AvgBytesPerSec / 1000);
this.outputBufferSize = (int)(this.bufferLengthInMs * outFormat.AvgBytesPerSec / 1000);
// Activate native Media Foundation COM objects on a thread-pool thread to ensure that they are in an MTA
Task.Run(() =>
{
DeviceUtil.CreateResamplerBuffer(this.inputBufferSize, out this.inputSample, out this.inputBuffer);
DeviceUtil.CreateResamplerBuffer(this.outputBufferSize, out this.outputSample, out this.outputBuffer);
// Create resampler object
this.resampler = DeviceUtil.CreateResampler(inFormat, outFormat);
}).Wait();
// Set the callback function
this.dataAvailableCallback = callback;
}
/// <summary>
/// Initialize the capturer.
/// </summary>
/// <param name="engineLatency">
/// Number of milliseconds of acceptable lag between live sound being produced and recording operation.
/// </param>
/// <param name="gain">
/// The gain to be applied to the audio after capture.
/// </param>
/// <param name="outFormat">
/// The format of the audio to be captured. If this is NULL, the default audio format of the
/// capture device will be used.
/// </param>
/// <param name="callback">
/// Callback function delegate which will handle the captured data.
/// </param>
/// <param name="speech">
/// If true, sets the audio category to speech to optimize audio pipeline for speech recognition.
/// </param>
public void Initialize(int engineLatency, float gain, WaveFormat outFormat, AudioDataAvailableCallback callback, bool speech)
{
// Create our shutdown event - we want a manual reset event that starts in the not-signaled state.
this.shutdownEvent = new ManualResetEvent(false);
// Now activate an IAudioClient object on our preferred endpoint and retrieve the mix format for that endpoint.
object obj = this.endpoint.Activate(ref audioClientIID, ClsCtx.INPROC_SERVER, IntPtr.Zero);
this.audioClient = (IAudioClient)obj;
// The following block enables advanced mic array APO pipeline on Windows 10 RS2 builds >= 15004.
// This must be called before the call to GetMixFormat() in LoadFormat().
if (speech)
{
IAudioClient2 audioClient2 = (IAudioClient2)this.audioClient;
if (audioClient2 != null)
{
AudioClientProperties properties = new AudioClientProperties
{
Size = Marshal.SizeOf <AudioClientProperties>(),
Category = AudioStreamCategory.Speech
};
int hr = audioClient2.SetClientProperties(ref properties);
if (hr != 0)
{
Console.WriteLine("Failed to set audio stream category to AudioCategory_Speech: {0}", hr);
}
}
else
{
Console.WriteLine("Unable to get IAudioClient2 interface");
}
}
// Load the MixFormat. This may differ depending on the shared mode used.
this.LoadFormat();
// Remember our configured latency
this.engineLatencyInMs = engineLatency;
// Set the gain
this.gain = gain;
// Determine whether or not we need a resampler
this.resampler = null;
if (outFormat != null)
{
// Check if the desired format is supported
IntPtr closestMatchPtr;
IntPtr outFormatPtr = WaveFormat.MarshalToPtr(outFormat);
int hr = this.audioClient.IsFormatSupported(AudioClientShareMode.Shared, outFormatPtr, out closestMatchPtr);
// Free outFormatPtr to prevent leaking memory
Marshal.FreeHGlobal(outFormatPtr);
if (hr == 0)
{
// Replace _MixFormat with outFormat. Since it is supported, we will initialize
// the audio capture client with that format and capture without resampling.
this.mixFormat = outFormat;
this.mixFrameSize = (this.mixFormat.BitsPerSample / 8) * this.mixFormat.Channels;
}
else
{
// In all other cases, we need to resample to OutFormat
if ((hr == 1) && (closestMatchPtr != IntPtr.Zero))
{
// Use closest match suggested by IsFormatSupported() and resample
this.mixFormat = WaveFormat.MarshalFromPtr(closestMatchPtr);
this.mixFrameSize = (this.mixFormat.BitsPerSample / 8) * this.mixFormat.Channels;
// Free closestMatchPtr to prevent leaking memory
Marshal.FreeCoTaskMem(closestMatchPtr);
}
this.inputBufferSize = (int)(this.engineLatencyInMs * this.mixFormat.AvgBytesPerSec / 1000);
this.outputBufferSize = (int)(this.engineLatencyInMs * outFormat.AvgBytesPerSec / 1000);
DeviceUtil.CreateResamplerBuffer(this.inputBufferSize, out this.inputSample, out this.inputBuffer);
DeviceUtil.CreateResamplerBuffer(this.outputBufferSize, out this.outputSample, out this.outputBuffer);
// Create resampler object
this.resampler = DeviceUtil.CreateResampler(this.mixFormat, outFormat);
}
}
this.InitializeAudioEngine();
// Set the callback function
this.dataAvailableCallback = callback;
}
