def main():
pairs = load_audio_path_label_pairs()
for index, (audio_path, _) in enumerate(pairs):
print('{} / {} ...'.format(index + 1, len(pairs)))
mg = compute_melgram(audio_path)
print('max: ', np.max(mg))
print('min: ', np.min(mg))
def main():
audio_path_label_pairs = load_audio_path_label_pairs()
shuffle(audio_path_label_pairs)
print('loaded: ', len(audio_path_label_pairs))
with tf.gfile.FastGFile('./models/tensorflow_models/cifar10/cifar10.pb', 'rb') as f:
graph_def = tf.GraphDef()
graph_def.ParseFromString(f.read())
_ = tf.import_graph_def(graph_def, name='')
with tf.Session() as sess:
[print(n.name) for n in sess.graph.as_graph_def().node]
predict_op = sess.graph.get_tensor_by_name('output_node0:0')
for i in range(0, 20):
audio_path, actual_label_id = audio_path_label_pairs[i]
mg = compute_melgram(audio_path)
mg = np.expand_dims(mg, axis=0)
predicted = sess.run(predict_op, feed_dict={"conv2d_1_input:0": mg})
predicted_label_idx = np.argmax(predicted, axis=1)[0]
predicted_label = gtzan_labels[predicted_label_idx]
actual_label = gtzan_labels[actual_label_id]
print('predicted: ', predicted_label, 'actual: ', actual_label)
def compute_melgram(self, audio_path):
if audio_path in self.cache:
return self.cache[audio_path]
else:
mg = compute_melgram(audio_path)
# mg = (mg + 100) / 200 # scale the values
self.cache[audio_path] = mg
return mg
def generate_batch(self, audio_paths, labels, batch_size):
num_batches = len(audio_paths) // batch_size
while True:
for batchIdx in range(0, num_batches):
start = batchIdx * batch_size
end = (batchIdx + 1) * batch_size
X = np.zeros(shape=(batch_size, self.input_shape[0], self.input_shape[1], self.input_shape[2]), dtype=np.float32)
for i in range(start, end):
audio_path = audio_paths[i]
mg = compute_melgram(audio_path)
X[i - start, :, :, :] = mg
yield X, labels[start:end]
def main():
audio_file_path = '../data/audio_samples/example.mp3'
# melgram_v1(audio_file_path, '../data/output/example_mp3.png')
# melgram_v2(audio_file_path)
arr = compute_melgram(audio_file_path)
print('melgram: ', arr.shape)
def predict(self, audio_path):
mg = compute_melgram(audio_path)
mg = np.expand_dims(mg, axis=0)
return self.model.predict(mg)[0]
