Update README.md

README.md

@@ -92,13 +92,14 @@ Sometimes it is useful to jointly visualize the VAD output with the input signal
 To do it:
 
 ```python
-upsampled_boundaries = VAD.upsample_boundaries(boundaries, audio_file)    
-torchaudio.save('vad_final.wav', upsampled_boundaries.cpu(), sample_rate) 
+import torchaudio
+upsampled_boundaries = VAD.upsample_boundaries(boundaries, 'pretrained_model_checkpoints/example_vad.wav')    
+torchaudio.save('vad_final.wav', upsampled_boundaries.cpu(), 16000) 
 ```  
 
 This creates a "VAD signal" with the same dimensionality as the original signal. 
 
-You can now open *vad_final.wav* and *speechbrain/vad_example.wav* with software like audacity to visualize them jointly. 
+You can now open *vad_final.wav* and *pretrained_model_checkpoints/example_vad.wav* with software like audacity to visualize them jointly. 
 
 
 ### VAD pipeline details
@@ -117,6 +118,7 @@ We designed the VAD such that you can have access to all of these steps (this mi
 ```python
 
 # 1- Let's compute frame-level posteriors first
+audio_file = 'pretrained_model_checkpoints/example_vad.wav'
 prob_chunks = VAD.get_speech_prob_file(audio_file)
 
 # 2- Let's apply a threshold on top of the posteriors