diff --git a/README.md b/README.md
index 30e2559..ec010d9 100644
--- a/README.md
+++ b/README.md
@@ -1,6 +1,12 @@
 # Spectral Amplitude Quantisation using NNs
 
-Experiments with time/frequency sample rate conversion and VQ-VAE for quantising the spectrum of speech for vocoders.
+Experiments with time/frequency sample rate conversion and VQ-VAE (Vector Quantised Variational Autoencoder) for quantising the speech spectrum for vocoders.
+
+This plot shows a VQ-VAE in action:
+
+![PCA of VQVAE encoder space](doc/vqvae_pca.png)
+
+The plot is a 2D histogram of the encoder space, white dots are the stage 1 VQ entries.  The 16 dimensional data has been reduced to 2 dimensions using PCA.  The plot was produced by `vq_vae_conv1d_2stage.py`
 
 ## Themes and Key Points
 
@@ -26,7 +32,7 @@ Experiments with time/frequency sample rate conversion and VQ-VAE for quantising
 | vq_vae_demo.py | Simple demo of VQ, nice visualisation of training in action | Cool demo | 
 | vq_vae_demo_2stage.py | vq_vae_demo.py extended to two stage VQ | Cool demo | 
 | vq_vae_ratek.py | Single stage VQ-VAE with single Dense layer | No |
-| vq_vae_ratek_conv1d.py | Two stage VQ-VAE with two conv1D layers | Yes, reasonable spectral distortion, cool plots |
+| vq_vae_conv1d_2stage.py | Two stage VQ-VAE with two conv1D layers | Yes, reasonable spectral distortion, cool plots |
 
 ## Amplitude Sample Rate Conversion Using Neural Nets
 
diff --git a/doc/vqvae_pca.png b/doc/vqvae_pca.png
new file mode 100644
index 0000000..998eff2
Binary files /dev/null and b/doc/vqvae_pca.png differ
diff --git a/vq_vae_conv1d_2stage.py b/vq_vae_conv1d_2stage.py
index 84c5ed7..59ae4fb 100755
--- a/vq_vae_conv1d_2stage.py
+++ b/vq_vae_conv1d_2stage.py
@@ -341,7 +341,7 @@ while key != 'q':
         plt.title(t)
     plt.show(block=False)
     plt.pause(0.0001)
-    print("n-next b-back s-save_png q-quit", end='', flush=True);
+    print("n-next b-back s-save_png q-quit", end='\r', flush=True);
     key = getch.getch()
     if key == 'n':
         fs += nb_plots