public GoogleSpeak(Pipeline pipeline, Microsoft.Psi.Audio.WaveFormat format, string languageCode) : base(pipeline)
{
this.gClient = TextToSpeechClient.Create();
this.format = format;
this.TextLanguageCode = languageCode;
if (format.Equals(Microsoft.Psi.Audio.WaveFormat.Create16kHz1Channel16BitPcm()))
{
this.googleAudioFormat = AudioEncoding.Linear16;
}
}
/// <summary>
/// Initializes a new instance of the <see cref="AudioBuffer"/> structure of a pre-allocated fixed length containing
/// no data initially. This overload is used primarily for making a copy of an existing <see cref="AudioBuffer"/>.
/// </summary>
/// <param name="length">The size in bytes of the audio data.</param>
/// <param name="format">The audio format.</param>
public AudioBuffer(int length, WaveFormat format)
: this(new byte[length], format)
{
}
/// <summary>
/// Initializes a new instance of the <see cref="AudioCapture"/> class with a specified output format and device name.
/// </summary>
/// <param name="pipeline">The pipeline to add the component to.</param>
/// <param name="outputFormat">The output format to use.</param>
/// <param name="deviceName">The name of the audio device.</param>
public AudioCapture(Pipeline pipeline, WaveFormat outputFormat, string deviceName = "plughw:0,0")
: this(pipeline, new AudioCaptureConfiguration() { Format = outputFormat, DeviceName = deviceName })
{
}
/// <summary>
/// Called when the pipeline is shutting down.
/// </summary>
private void OnPipelineStop()
{
this.sourceFormat = null;
}
/// <summary>
/// Initializes a new instance of the <see cref="AudioBuffer"/> structure.
/// </summary>
/// <param name="data">An array of bytes containing the audio data.</param>
/// <param name="format">The audio format.</param>
public AudioBuffer(byte[] data, WaveFormat format)
{
this.data = data;
this.format = format;
}
/// <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="engineBuffer">
/// Number of milliseconds of audio that may be buffered between reads.
/// </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, int engineBuffer, 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 and buffer size
this.engineLatencyInMs = engineLatency;
this.engineBufferInMs = engineBuffer;
// 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;
this.InitializeAudioEngine();
}
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);
}
// initialize the audio engine first as the engine latency may be modified after initialization
this.InitializeAudioEngine();
// initialize the resampler buffers
this.inputBufferSize = (int)(this.engineBufferInMs * this.mixFormat.AvgBytesPerSec / 1000);
this.outputBufferSize = (int)(this.engineBufferInMs * 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);
}
}
else
{
// initialize the audio engine with the default mix format
this.InitializeAudioEngine();
}
// Set the callback function
this.dataAvailableCallback = callback;
}
/// <summary>
/// Resamples an audio stream.
/// </summary>
/// <param name="source">A stream audio to be resampled.</param>
/// <param name="outputFormat">The desired audio output format for the resampled stream.</param>
/// <returns>A stream of resampled audio.</returns>
public static IProducer <AudioBuffer> Resample(this IProducer <AudioBuffer> source, WaveFormat outputFormat)
{
return(Resample(source, new AudioResamplerConfiguration()
{
OutputFormat = outputFormat
}));
}
/// <summary>
/// Initializes a new instance of the <see cref="AudioCapture"/> class with a specified output format and device name.
/// </summary>
/// <param name="pipeline">The pipeline to add the component to.</param>
/// <param name="outputFormat">The output format to use.</param>
/// <param name="deviceName">The name of the audio device.</param>
/// <param name="name">An optional name for the component.</param>
public AudioCapture(Pipeline pipeline, WaveFormat outputFormat, string deviceName = null, string name = nameof(AudioCapture))
: this(pipeline, new AudioCaptureConfiguration() { Format = outputFormat, DeviceName = deviceName }, name)
{
}
/// <inheritdoc/>
public void Stop(DateTime finalOriginatingTime, Action notifyCompleted)
{
notifyCompleted();
this.sourceFormat = null;
}
/// <summary>
/// Returns the size in bytes required to create a marshaled unmanaged copy of the object.
/// </summary>
/// <param name="format">The object that is to be marshaled.</param>
/// <returns>The unmanaged size of the object.</returns>
public static int MarshalSizeOf(WaveFormat format)
{
return(Marshal.SizeOf <WaveFormat>() + format?.ExtraSize ?? 0);
}
/// <summary>
/// Starts rendering audio data.
/// </summary>
/// <param name="maxBufferSeconds">
/// The maximum duration of audio that can be buffered for playback.
/// </param>
/// <param name="targetLatencyInMs">
/// The target maximum number of milliseconds of acceptable lag between
/// playback of samples and live sound being produced.
/// </param>
/// <param name="gain">
/// The gain to be applied prior to rendering the audio.
/// </param>
/// <param name="inFormat">
/// The input audio format.
/// </param>
public void StartRendering(double maxBufferSeconds, int targetLatencyInMs, float gain, WaveFormat inFormat)
{
if (this.wasapiRenderClient != null)
{
this.StopRendering();
}
// Create an audio buffer to buffer audio awaiting playback.
this.audioBufferStream = new CircularBufferStream((long)Math.Ceiling(maxBufferSeconds * inFormat.AvgBytesPerSec), false);
this.wasapiRenderClient = new WasapiRenderClient(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 AudioDataRequestedCallback(this.AudioDataRequestedCallback);
// initialize the renderer with the desired parameters
this.wasapiRenderClient.Initialize(targetLatencyInMs, gain, inFormat, this.callbackDelegate);
// tell WASAPI to start rendering
this.wasapiRenderClient.Start();
}
/// <summary>
/// Converts a stream of audio data to a stream of floating point values.
/// </summary>
/// <param name="source">A stream containing the input audio data.</param>
/// <param name="format">The audio format of the input audio.</param>
/// <param name="deliveryPolicy">An optional delivery policy.</param>
/// <returns>A stream of floating point audio sample values.</returns>
public static IProducer <float[]> ToFloat(this IProducer <byte[]> source, WaveFormat format, DeliveryPolicy deliveryPolicy = null)
{
return(source.PipeTo(new ToFloat(source.Out.Pipeline, format), deliveryPolicy));
}
/// <summary>
/// Transforms a stream of byte arrays containing raw audio to an <see cref="AudioBuffer"/> stream.
/// </summary>
/// <param name="source">A stream of raw audio byte arrays.</param>
/// <param name="audioFormat">The audio format of the raw audio contained within the byte arrays.</param>
/// /// <param name="deliveryPolicy">An optional delivery policy.</param>
/// <returns>A stream of audio buffers.</returns>
public static IProducer <AudioBuffer> ToAudioBuffer(this IProducer <byte[]> source, WaveFormat audioFormat, DeliveryPolicy deliveryPolicy = null)
{
return(source.Select(x => new AudioBuffer(x, audioFormat), deliveryPolicy));
}
/// <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="audioEngineBufferInMs">
/// The amount of audio that may be buffered by the audio engine between
/// reads.
/// </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>
/// <param name="eventDrivenCapture">
/// If true, initialize Windows audio capture in event-driven mode. The audio capture engine will call
/// the <see cref="AudioDataAvailableCallback"/> handler as soon as data is available, at intervals
/// determined by the audio engine (which may be less than the <paramref name="targetLatencyInMs"/>).
/// This captures audio with the lowest possible latency while still allowing for buffering up to the
/// amount of time specified by <paramref name="targetLatencyInMs"/> (when for example the system is
/// under heavy load and the capture callback is unable to service audio packets at the rate at which
/// the audio engine returns captured audio packets).
/// </param>
public void StartCapture(int targetLatencyInMs, int audioEngineBufferInMs, float gain, WaveFormat outFormat, bool speech, bool eventDrivenCapture)
{
if (this.wasapiCaptureClient != null)
{
this.StopCapture();
}
this.wasapiCaptureClient = new WasapiCaptureClient(this.audioDevice, eventDrivenCapture);
// 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.wasapiCaptureClient.Initialize(targetLatencyInMs, audioEngineBufferInMs, gain, outFormat, this.callbackDelegate, speech);
// tell WASAPI to start capturing
this.wasapiCaptureClient.Start();
}