Cannot batch tensors with different shapes in component 0. First element had shape [1445,40] and element 1 had shape [1330,40].

[himrlawrrence]

==>
hi vinneths96,

i tried to use my own hotword -"get" insteand of "marvin" .
Then added some get-wav files into train folder ,create_modle and when run
"
# Obtain the feature embeddings
X_train = feature_extractor.predict(get_data, use_multiprocessing=True)
"
i got the err stack:
Cannot batch tensors with different shapes in component 0. First element had shape [1445,40] and element 1 had shape [1330,40].

could you please help me out?

Thanks a lot.

JFU

[vineeths96]

Hi JFU,

Can you post the entire error stack to understand which statement triggers this error? It's hard to understand from the current short snapshot you provided.

[himrlawrrence]

i'm glad to get ur quick reply.
i have another quick question: trained wav files must be the same size, eg.32K OR, must be less than 1 second?

[himrlawrrence]

error stack is shown below:

C:\Users\JFU\anaconda3\envs\env38\python.exe C:/WORKSPACE/Spoken-Keyword-Spotting/src/main.py
2021-11-02 13:58:01.704671: W tensorflow/stream_executor/platform/default/dso_loader.cc:55] Could not load dynamic library 'cudart64_101.dll'; dlerror: cudart64_101.dll not found
2021-11-02 13:58:01.704822: I tensorflow/stream_executor/cuda/cudart_stub.cc:29] Ignore above cudart dlerror if you do not have a GPU set up on your machine.
Training model
Dataset statistics
Train files: 51410
Validation files: 6640
Dev test files: 6675
Test files: 2567
2021-11-02 13:58:07.875339: W tensorflow/stream_executor/platform/default/dso_loader.cc:55] Could not load dynamic library 'nvcuda.dll'; dlerror: nvcuda.dll not found
2021-11-02 13:58:07.875475: E tensorflow/stream_executor/cuda/cuda_driver.cc:313] failed call to cuInit: UNKNOWN ERROR (303)
2021-11-02 13:58:07.881674: I tensorflow/stream_executor/cuda/cuda_diagnostics.cc:169] retrieving CUDA diagnostic information for host: JFU-LAPTOP
2021-11-02 13:58:07.881881: I tensorflow/stream_executor/cuda/cuda_diagnostics.cc:176] hostname: JFU-LAPTOP
2021-11-02 13:58:07.882268: I tensorflow/core/platform/cpu_feature_guard.cc:143] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX2
2021-11-02 13:58:07.891825: I tensorflow/compiler/xla/service/service.cc:168] XLA service 0x19452a35900 initialized for platform Host (this does not guarantee that XLA will be used). Devices:
2021-11-02 13:58:07.891975: I tensorflow/compiler/xla/service/service.cc:176] StreamExecutor device (0): Host, Default Version
Model: "sequential"
...
...
..
Total params: 930,403
Trainable params: 927,873
Non-trainable params: 2,530

---

Epoch 1/25
401/401 [==============================] - 221s 550ms/step - loss: 2.2773 - sparse_categorical_accuracy: 0.3448 - val_loss: 1.2852 - val_sparse_categorical_accuracy: 0.6045 - lr: 0.0010
Epoch 2/25
401/401 [==============================] - 269s 672ms/step - loss: 0.8611 - sparse_categorical_accuracy: 0.7339 - val_loss: 0.5980 - val_sparse_categorical_accuracy: 0.8114 - lr: 0.0010
Epoch 3/25
401/401 [==============================] - 290s 724ms/step - loss: 0.5668 - sparse_categorical_accuracy: 0.8260 - val_loss: 0.3616 - val_sparse_categorical_accuracy: 0.8905 - lr: 0.0010
.........
Epoch 23/25
401/401 [==============================] - 404s 1s/step - loss: 0.1639 - sparse_categorical_accuracy: 0.9494 - val_loss: 0.1798 - val_sparse_categorical_accuracy: 0.9487 - lr: 0.0010
Epoch 24/25
401/401 [==============================] - 408s 1s/step - loss: 0.1644 - sparse_categorical_accuracy: 0.9489 - val_loss: 0.1863 - val_sparse_categorical_accuracy: 0.9490 - lr: 0.0010
Epoch 25/25
401/401 [==============================] - 420s 1s/step - loss: 0.1572 - sparse_categorical_accuracy: 0.9519 - val_loss: 0.1741 - val_sparse_categorical_accuracy: 0.9487 - lr: 0.0010
Saving model
Saving training history
Traceback (most recent call last):
File "C:/WORKSPACE/Spoken-Keyword-Spotting/src/main.py", line 25, in
main()
File "C:/WORKSPACE/Spoken-Keyword-Spotting/src/main.py", line 21, in main
get_kws_model()
File "C:\WORKSPACE\Spoken-Keyword-Spotting\src\model_train.py", line 141, in get_kws_model
X_train = feature_extractor.predict(get_data, use_multiprocessing=True)
File "C:\Users\JFU\anaconda3\envs\env38\lib\site-packages\tensorflow\python\keras\engine\training.py", line 88, in _method_wrapper
return method(self, *args, **kwargs)
File "C:\Users\JFU\anaconda3\envs\env38\lib\site-packages\tensorflow\python\keras\engine\training.py", line 1268, in predict
tmp_batch_outputs = predict_function(iterator)
File "C:\Users\JFU\anaconda3\envs\env38\lib\site-packages\tensorflow\python\eager\def_function.py", line 580, in call
result = self._call(*args, **kwds)
File "C:\Users\JFU\anaconda3\envs\env38\lib\site-packages\tensorflow\python\eager\def_function.py", line 650, in _call
return self._concrete_stateful_fn._filtered_call(canon_args, canon_kwds) # pylint: disable=protected-access
File "C:\Users\JFU\anaconda3\envs\env38\lib\site-packages\tensorflow\python\eager\function.py", line 1661, in _filtered_call
return self._call_flat(
File "C:\Users\JFU\anaconda3\envs\env38\lib\site-packages\tensorflow\python\eager\function.py", line 1745, in _call_flat
return self._build_call_outputs(self._inference_function.call(
File "C:\Users\JFU\anaconda3\envs\env38\lib\site-packages\tensorflow\python\eager\function.py", line 593, in call
outputs = execute.execute(
File "C:\Users\JFU\anaconda3\envs\env38\lib\site-packages\tensorflow\python\eager\execute.py", line 59, in quick_execute
tensors = pywrap_tfe.TFE_Py_Execute(ctx._handle, device_name, op_name,
tensorflow.python.framework.errors_impl.InvalidArgumentError: Cannot batch tensors with different shapes in component 0. First element had shape [1445,40] and element 1 had shape [1330,40].
[[node IteratorGetNext (defined at \WORKSPACE\Spoken-Keyword-Spotting\src\model_train.py:141) ]] [Op:__inference_predict_function_29742]

Function call stack:
predict_function

Process finished with exit code 1

[vineeths96]

I believe the sampling rate should be the same as the one used for training.

[himrlawrrence]

hi vinneth,
how to train other key word?
thanks,
JFU