This repository has been archived by the owner on Jul 17, 2024. It is now read-only.
Issues with "Training an audio keyword spotter with PyTorch" tutorial #232
Comments
|
Good point. The pyaudio setup was mentioned in the first audio tutorial. ONNX setup is missing from all our docs. Thanks for reporting these doc bugs. |
lovettchris
added a commit
that referenced
this issue
Dec 29, 2019
… tutorial Fix github issue #232 - regressions in audio training tutorial, some command line args have changed, added a blub on VAD tuning that seems to be needed based on some email conversations with some users.
kernhanda
pushed a commit
that referenced
this issue
Jun 10, 2020
ELL release v3.1.0 - Move to VS 2019 - Fix a codegen error that was resulting in incorrect functional behavior - Fix regressions in audio training tutorial (#232) - Add importing of Sum nodes to ONNX importer - Fix crash in LLVMContext::SetName - Improved performance of CNN models on Pi3 with new implementations of spatial, pointwise and regular convolutions - Improved performance of reorder node - New nodes: ReorderDataCodeNode, SpatialConvolutionNode, MatrixMatrixMultiplyCodeNode - Implement parallelization strategies for matrix multiplication nodes. - Only enable new MatrixMatrixMultipleCodeNode path for select ARM targets like Pi, and not Intel/AMD CPUs - Add the flag `--skip_ellcode` to `compile` and `wrap.py` tools to use OpenBLAS for linear algebra computations.
|
Most of these issues are fixed and "onnx" is listed in the requirements.txt file in the ELL repo. |
Sign up for free
to subscribe to this conversation on GitHub.
Already have an account?
Sign in.
I have followed the Training an audio keyword spotter with PyTorch tutorial, on a clean install of Ubuntu 18.04.
I have found the following minor issues:
pyaudio.onnx- I used the following commandconda install -c conda-forge onnx=1.4.1The text was updated successfully, but these errors were encountered: