add SNORING_INDEX

python/util/plt_util.py

@@ -28,6 +28,7 @@ def plt_line(y_points, sample_rate=16000):
     fig, ax = plt.subplots()
     ax.xaxis.set_major_formatter(ticker.FuncFormatter(update_ticks))
     plt.plot(y_points)
+    plt.title('Waveform')
 
     # plot to image
     buffer = BytesIO()
@@ -44,7 +45,7 @@ def plt_mfcc(single_channel, sample_rate):
     plt.figure()
     librosa.display.specshow(log_mel_spec, sr=sample_rate, x_axis='time', y_axis='mel')
     plt.colorbar(format='%+2.0f dB')  # 右边的色度条
-    plt.title('mfcc waveform')
+    plt.title('MFCC')
 
     # plot to image
     buffer = BytesIO()

test.py

@@ -20,6 +20,8 @@ OUT_PCM = 'PCM_16'
 CLASS_MAP_FILE = 'res/yamnet_class_map.csv'
 DEBUG = True
 SNORING_TOP_N = 7
+SNORING_INDEX = 38
+IN_MODEL_SAMPLES = 15600
 
 
 # Methods
@@ -81,7 +83,7 @@ def predict_waveform(idx, waveform):
         score = means[index]
         name = class_names[index]
 
-        if name == 'Snoring':
+        if index == SNORING_INDEX:
             snoring_score = score
         top_n_res += ' ' + format_float(score) + ' [' + truncate_str(name, 4) + '], '
 
@@ -112,7 +114,7 @@ def predict_uri(audio_uri1, audio_uri2):
     wav_input = audio_to_wav(mp3_input) if not mp3_input.endswith('.mp3') == True else mp3_input
     predict_seconds = int(str(sys.argv[2])) if len(sys.argv) > 2 else 1
 
-    predict_samples = 15600 #OUT_SAMPLE_RATE * predict_seconds
+    predict_samples = IN_MODEL_SAMPLES  # OUT_SAMPLE_RATE * predict_seconds
     single_channel, sc_sample_rate = read_single_channel(wav_input)
     splits = split_given_size(single_channel, predict_samples)
     result += ' sc_sample_rate: ' + str(sc_sample_rate) + '\n'