Define DSP::Reset() and DSP:Warmup()

NAM/dsp.cpp

@@ -52,6 +52,14 @@ double nam::DSP::GetLoudness() const
   return mLoudness;
 }
 
+void nam::DSP::Reset(const double sampleRate, const int maxBufferSize)
+{
+  // Some subclasses might want to throw an exception if the sample rate is "wrong".
+  // This could be under a debugging flag potentially.
+  mExternalSampleRate = sampleRate;
+  mHaveExternalSampleRate = true;
+  mMaxBufferSize = maxBufferSize;
+}
 void nam::DSP::SetLoudness(const double loudness)
 {
   mLoudness = loudness;

NAM/dsp.h

@@ -64,10 +64,16 @@ class DSP
   double GetLoudness() const;
   // Get whether the model knows how loud it is.
   bool HasLoudness() const { return mHasLoudness; };
+  // General function for resetting the DSP unit.
+  // NAMs might do warmup under this.
+  virtual void Reset(const double sampleRate, const int maxBufferSize);
   // Set the loudness, in dB.
   // This is usually defined to be the loudness to a standardized input. The trainer has its own, but you can always
   // use this to define it a different way if you like yours better.
   void SetLoudness(const double loudness);
+  // Run some zeroes through the DSP unit until it's ready.
+  // This is helpful for things that have a history dependence (LSTMs, Convolutions, etc)
+  void Warmup();
 
 protected:
   bool mHasLoudness = false;
@@ -77,6 +83,14 @@ class DSP
   double mExpectedSampleRate;
   // How many samples should be processed during "pre-warming"
   int _prewarm_samples = 0;
+  // Have we been told what the external sample rate is? If so, what is it?
+  bool mHaveExternalSampleRate = false;
+  double mExternalSampleRate = -1.0;
+  // The largest buffer I expect to be told to process:
+  int mMaxBufferSize;
+
+  // How many samples should be processed for me to be considered "warmed up"?
+  virtual int WarmupSamples() { return 0; };
 };
 
 // Class where an input buffer is kept so that long-time effects can be
@@ -87,6 +101,8 @@ class Buffer : public DSP
 public:
   Buffer(const int receptive_field, const double expected_sample_rate = -1.0);
   void finalize_(const int num_frames);
+  // TODO Could set buffer sizes etc
+  // virtual void Reset(const double sampleRate, const int maxBufferSize) override;
 
 protected:
   // Input buffer
@@ -120,6 +136,7 @@ class Linear : public Buffer
 
 // NN modules =================================================================
 
+// TODO conv could take care of its own ring buffer.
 class Conv1D
 {
 public:

NAM/wavenet.cpp

@@ -30,7 +30,6 @@ void nam::wavenet::_Layer::process_(const Eigen::MatrixXf& input, const Eigen::M
   // Mix-in condition
   this->_z += this->_input_mixin.process(condition);
 
-
 
   if (!this->_gated)
   {
@@ -40,7 +39,8 @@ void nam::wavenet::_Layer::process_(const Eigen::MatrixXf& input, const Eigen::M
   {
     this->_activation->apply(this->_z.topRows(channels));
     activations::Activation::get_activation("Sigmoid")->apply(this->_z.bottomRows(channels));
-    //activations::Activation::get_activation("Sigmoid")->apply(this->_z.block(channels, 0, channels, this->_z.cols()));
+    // activations::Activation::get_activation("Sigmoid")->apply(this->_z.block(channels, 0, channels,
+    // this->_z.cols()));
 
     this->_z.topRows(channels).array() *= this->_z.bottomRows(channels).array();
     // this->_z.topRows(channels) = this->_z.topRows(channels).cwiseProduct(