OSSForks/tensorflow
1
0
Fork 0
mirror of https://github.com/zebrajr/tensorflow.git synced 2025-12-06 12:20:11 +01:00
Code Issues Actions 5 Packages Projects Releases Wiki Activity

Fixed input shape for freezing audio graphs

Browse Source 
PiperOrigin-RevId: 165649546
This commit is contained in:
Pete Warden 2017-08-17 17:05:12 -07:00 committed by TensorFlower Gardener
parent 9b9e5989d2
commit a3c4e980e0
1 changed files with 6 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

7
tensorflow/examples/speech_commands/freeze.py
View File

@ -90,9 +90,14 @@ def create_inference_graph(wanted_words, sample_rate, clip_duration_ms,
spectrogram,
decoded_sample_data.sample_rate,
dct_coefficient_count=dct_coefficient_count)
fingerprint_frequency_size = model_settings['dct_coefficient_count']
fingerprint_time_size = model_settings['spectrogram_length']
reshaped_input = tf.reshape(fingerprint_input, [
-1, fingerprint_time_size * fingerprint_frequency_size
])
logits = models.create_model(
fingerprint_input, model_settings, model_architecture, is_training=False)
reshaped_input, model_settings, model_architecture, is_training=False)
# Create an output to use for inference.
tf.nn.softmax(logits, name='labels_softmax')