/// <summary>
/// Gets a list of available audio capture devices.
/// </summary>
/// <returns>
/// An array of available capture device names.
/// </returns>
public static string[] GetAvailableCaptureDevices()
{
// Get the collection of available capture devices
IMMDeviceCollection deviceCollection = DeviceUtil.GetAvailableDevices(EDataFlow.Capture);
string[] devices = null;
int deviceCount = deviceCollection.GetCount();
devices = new string[deviceCount];
// Iterate over the collection to get the device names
for (int i = 0; i < deviceCount; i++)
{
IMMDevice device = deviceCollection.Item(i);
// Get the friendly name of the device
devices[i] = DeviceUtil.GetDeviceFriendlyName(device);
// Done with the device so release it
Marshal.ReleaseComObject(device);
}
// Release the collection when done
Marshal.ReleaseComObject(deviceCollection);
return(devices);
}
/// <summary>
/// Initializes the audio capture device.
/// </summary>
/// <param name="deviceDescription">
/// The friendly name description of the device to capture from. This is usually
/// something like "Microphone Array (USB Audio)". To capture from
/// the default device, pass in NULL or an empty string.
/// </param>
public void Initialize(string deviceDescription)
{
// Activate native audio COM objects on a thread-pool thread to ensure that they are in an MTA
Task.Run(() =>
{
if (string.IsNullOrEmpty(deviceDescription))
{
// use the default console device
this.audioDevice = DeviceUtil.GetDefaultDevice(EDataFlow.Capture, ERole.Console);
}
else
{
this.audioDevice = DeviceUtil.GetDeviceByName(EDataFlow.Capture, deviceDescription);
}
if (this.audioDevice != null)
{
// Try to get the volume control
object obj = this.audioDevice.Activate(new Guid(Guids.IAudioEndpointVolumeIIDString), ClsCtx.ALL, IntPtr.Zero);
this.volume = (IAudioEndpointVolume)obj;
// Now create an IAudioEndpointVolumeCallback object that wraps the callback and register it with the endpoint.
this.volumeCallback = new AudioEndpointVolumeCallback(this.AudioVolumeCallback);
this.volume.RegisterControlChangeNotify(this.volumeCallback);
}
}).Wait();
}
/// <summary>
/// Initializes the audio capture device.
/// </summary>
/// <param name="deviceDescription">
/// The friendly name description of the device to capture from. This is usually
/// something like "Microphone Array (USB Audio)". To capture from
/// the default device, pass in NULL or an empty string.
/// </param>
public void Initialize(string deviceDescription)
{
Exception taskException = null;
// Activate native audio COM objects on a thread-pool thread to ensure that they are in an MTA
Task.Run(() =>
{
try
{
if (string.IsNullOrEmpty(deviceDescription))
{
// use the default console device
this.audioDevice = DeviceUtil.GetDefaultDevice(EDataFlow.Capture, ERole.Console);
}
else
{
this.audioDevice = DeviceUtil.GetDeviceByName(EDataFlow.Capture, deviceDescription);
}
if (this.audioDevice != null)
{
// Try to get the volume control
object obj = this.audioDevice.Activate(new Guid(Guids.IAudioEndpointVolumeIIDString), ClsCtx.ALL, IntPtr.Zero);
this.volume = (IAudioEndpointVolume)obj;
// Now create an IAudioEndpointVolumeCallback object that wraps the callback and register it with the endpoint.
this.volumeCallback = new AudioEndpointVolumeCallback(this.AudioVolumeCallback);
this.volume.RegisterControlChangeNotify(this.volumeCallback);
}
}
catch (Exception e)
{
taskException = e;
}
}).Wait();
// do error checking on the main thread
if (taskException != null)
{
// rethrow exception
throw taskException;
}
else if (this.audioDevice == null)
{
throw new IOException(string.IsNullOrEmpty(deviceDescription) ?
"No default audio capture device found." :
$"Audio capture device {deviceDescription} not found.");
}
}
/// <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;
}
/// <summary>
/// Initialize the renderer.
/// </summary>
/// <param name="engineLatency">
/// Number of milliseconds of acceptable lag between playback of samples and live sound being produced.
/// </param>
/// <param name="gain">
/// The gain to be applied to the audio before rendering.
/// </param>
/// <param name="inFormat">
/// The format of the input audio samples to be rendered. If this is NULL, the current default audio
/// format of the renderer device will be assumed.
/// </param>
/// <param name="callback">
/// Callback function delegate which will supply the data to be rendered.
/// </param>
public void Initialize(int engineLatency, float gain, WaveFormat inFormat, AudioDataRequestedCallback callback)
{
// 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;
// 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;
// Check if the desired format is supported
IntPtr closestMatchPtr;
IntPtr inFormatPtr = WaveFormat.MarshalToPtr(inFormat);
int hr = this.audioClient.IsFormatSupported(AudioClientShareMode.Shared, inFormatPtr, out closestMatchPtr);
// Free outFormatPtr to prevent leaking memory
Marshal.FreeHGlobal(inFormatPtr);
if (hr == 0)
{
// Replace _MixFormat with inFormat. Since it is supported, we will initialize
// the audio render client with that format and render without resampling.
this.mixFormat = inFormat;
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 * inFormat.AvgBytesPerSec / 1000);
this.outputBufferSize = (int)(this.engineLatencyInMs * this.mixFormat.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(inFormat, this.mixFormat);
this.InitializeAudioEngine();
// Set the callback function
this.dataRequestedCallback = callback;
}
