Top AI Repos — open-source AI, indexed and scored
Top AI Repos tracks AI repositories on GitHub and answers two different questions about each one: is it moving right now, and would you bet a product on it.
Top AI Repos tracks AI repositories on GitHub and answers two different questions about each one: is it moving right now, and would you bet a product on it.
Speech-to-Text-WaveNet : End-to-end sentence level English speech recognition based on DeepMind's WaveNet and tensorflow
| Date | Stars |
|---|---|
| 2026-07-31 | 4005 |
| 2026-08-01 | 4005 |
| 2026-08-03 | 4005 |
| 2026-08-06 | 4005 |
Today
— stars today
This week
— stars this week
This month
— stars this month
Momentum
0.0
growth rate 0.00%/day
# Speech-to-Text-WaveNet : End-to-end sentence level English speech recognition using DeepMind's WaveNet
A tensorflow implementation of speech recognition based on DeepMind's [WaveNet: A Generative Model for Raw Audio](https://arxiv.org/abs/1609.03499). (Hereafter the Paper)
Although [ibab](https://github.com/ibab/tensorflow-wavenet) and [tomlepaine](https://github.com/tomlepaine/fast-wavenet) have already implemented WaveNet with tensorflow, they did not implement speech recognition. That's why we decided to implement it ourselves.
Some of Deepmind's recent papers are tricky to reproduce. The Paper also omitted specific details about the implementation, and we had to fill the gaps in our own way.
Here are a few important notes.
First, while the Paper used the TIMIT dataset for the speech recognition experiment, we used the free VTCK dataset.
Second, the Paper added a mean-pooling layer after the dilated convolution layer for down-sampling. We extracted [MFCC](https://en.wikipedia.org/wiki/Mel-frequency_cepstrum) from wav files and removed the final mean-pooling layer because the original setting was impossible to run on our TitanX GPU.
Third, since the TIMIT dataset has phoneme labels, the Paper trained the model with two loss terms, phoneme classification and next phoneme prediction. We, instead, used a single CTC loss because VCTK provides sentence-level labels. As a result, we used only dilated conv1d layers without any dilated conv1d layers.
Finally, we didn't do quantitative analyses such as BLEU score and post-processing by combining a language model due to the time constraints.
The final architecture is shown in the following figure.
<p align="center">
<img src="https://raw.githubusercontent.com/buriburisuri/speech-to-text-wavenet/master/png/architecture.png" width="1024"/>
</p>
(Some images are cropped from [WaveNet: A Generative Model for Raw Audio](https://arxiv.org/abs/1609.03499) and [Neural Machine Translation in Linear Time](https://arxiv.org/abs/1610.10099))
## Version
Current Version : __***0.0.0.2***__
## Dependencies ( VERSION MUST BE MATCHED EXACTLY! )
1. [tensorflow](https://www.tensorflow.org/versions/r0.11/get_started/os_setup.html#pip-installation) == 1.0.0
1. [sugartensor](https://github.com/buriburisuri/sugartensor) == 1.0.0.2
1. [pandas](http://pandas.pydata.org/pandas-docs/stable/install.html) >= 0.19.2
1. [librosa](https://github.com/librosa/librosa) == 0.5.0
1. [scikits.audiolab](https://pypi.python.org/pypi/scikits.audiolab)==0.11.0
If you have problems with the librosa library, try to install ffmpeg by the following command. ( Ubuntu 14.04 )
<pre><code>
sudo add-apt-repository ppa:mc3man/trusty-media
sudo apt-get update
sudo apt-get dist-upgrade -y
sudo apt-get -y install ffmpeg
</code></pre>
## Dataset
We used [VCTK](http://homepages.inf.ed.ac.uk/jyamagis/page3/page58/page58.html),
[LibriSpeech](http://www.openslr.org/12/) and [TEDLIUM release 2](http://www-lium.univ-lemans.fr/en/content/ted-lium-corpus) corpus.
Total number of sentences in the training set composed of the above three corpus is 240,612.
Valid and test set is built using only LibriSpeech and TEDLIUM corpuse, because VCTK corpus does not have valid and test set.
After downloading the each corpus, extract them in the 'asset/data/VCTK-Corpus', 'asset/data/LibriSpeech' and
'asset/data/TEDLIUM_release2' directories.
Audio was augmented by the scheme in the [Tom Ko et al](http://speak.clsp.jhu.edu/uploads/publications/papers/1050_pdf.pdf)'s paper.
(Thanks @migvel for your kind information)
## Pre-processing dataset
The TEDLIUM release 2 dataset provides audio data in the SPH format, so we should convert them to some format
librosa library can handle. Run the following command in the 'asset/data' directory convert SPH to wave format.
<pre><code>
find -type f -name '*.sph' | awk '{printf "sox -t sph %s -b 16 -t wav %s\n", $0, $0".wav" }' | bash
</code></pre>
If you don't have installed `sox`, pleExcerpt of 9,279 characters
Read on GitHub
15
Mike C. Fletcher · VRPlumber Consulting Inc. · Canada
1
1
1
Would you bet a product on this? Bounded 0–100 and slow moving.
matched fp:53a039e066d3330c, desc:speech recognition, name:speech-to-text, desc:speech-to-text