MLSpeech · aguerrerolopez · Dec 27, 2023 · Dec 27, 2023 · Dec 27, 2023 · Dec 29, 2023
diff --git a/.gitignore b/.gitignore
@@ -0,0 +1,3 @@
+/__pycache__
+
+/**/Contr*
diff --git a/conf/config.yaml b/conf/config.yaml
@@ -26,6 +26,10 @@ gaussian_kernel_sigma: 1
 is_cuda: true
 test_batch_size: 4
 
+# Plot
+plot_formants: true
+plot_spectrogram: true
+
 
 # model
 f1_blocks: 2

diff --git a/dataloader.py b/dataloader.py
@@ -20,6 +20,13 @@ def preemphasis(x, coeff=0.97):
 
 def extract_features(wav_file, hp):
     wav, sr = taudio.load(wav_file)
+    # If SR is not 16kHz, resample
+    if sr != hp.sample_rate:
+        print(f"Resampling {wav_file} from {sr} to {hp.sample_rate}")
+        wav = taudio.transforms.Resample(sr, hp.sample_rate)(wav)
+        sr = hp.sample_rate
+
+    # Pre-emphasis
     if hp.emph>0:
         wav=preemphasis(wav,coeff=hp.emph)
 

diff --git a/predictions/Tonsill_ses3_speech_0007.png b/predictions/Tonsill_ses3_speech_0007.png