audiomate
audiomate copied to clipboard
Python library for handling audio datasets.
Calling from german-asr [local/prepare_data.py](https://github.com/german-asr/kaldi-german/blob/master/local/prepare_data.py) it looks like the process aborts after 3 hours consistently, without any diagnostics. Just the `train/wav.scp` is written, and that happens in the first few minutes....
We're currently only building and running the test suite on Linux. We should at least include macOS and, if possible, Windows. Question is what service we want to use. Travis...
If an utterance has no issuer an exception is thrown while converting. https://github.com/ynop/audiomate/blob/8ce371b816c02a401ce6d348c5c2d8de7566509a/audiomate/corpus/conversion/base.py#L56
- [Keras](https://keras.io) - [MXNet](https://mxnet.apache.org) - [TensorFlow](http://tensorflow.org) - [Caffe2](https://caffe2.ai)
I am trying to build the latest version of audiomate for openSUSE Linux and I am getting the following errors on 32bit builds but not 64bit ones: ``` =================================== FAILURES...
This issue is intended to store invalid utterances until they are incorporated into the code. SWC: ``` 00000394_0000000721_158260_159270 (Neugersdorf/audio.ogg 158.260 159.270) ```
Set-up a self-written GitHub action that checks URLs of datasets in code and documentation #88
The urls used in the code and documentation should be checked for availability (As done for the readme in #87) Places to check: * URLs in the CorpusDownloader classes *...
Hello, looking for some direction for my first dataset contribution. I've copied the repo to my local computer and have installed the environment.
The names of readers/writers are only listed in the source-code. (The name that is needed in ```Corpus.load(..., reader=[name])```) A list should be available in the documentation. e.g. in the list...