Repository files navigation

Future wav2letter development will occur in Flashlight.

To build the old, pre-consolidation version of wav2letter, checkout thewav2letter v0.2 release, which depends on the oldFlashlight v0.2 release. Thewav2letter-lua project can be found on thewav2letter-lua branch, accordingly.

For more information on wav2letter++, see or citethis arXiv paper.

This repository includes recipes to reproduce the following research papers as well aspre-trained models.All results reproduction must use Flashlight <= 0.3.2 for exact reproducability. Papers contained here include:

• Pratap et al. (2020): Scaling Online Speech Recognition Using ConvNets
• Synnaeve et al. (2020): End-to-end ASR: from Supervised to Semi-Supervised Learning with Modern Architectures
• Kahn et al. (2020): Self-Training for End-to-End Speech Recognition
• Likhomanenko et al. (2019): Who Needs Words? Lexicon-free Speech Recognition
• Hannun et al. (2019): Sequence-to-Sequence Speech Recognition with Time-Depth Separable Convolutions

Data preparation for training and evaluation can be found indata directory.

First, installFlashlight(using the0.3 branch is required) with theASR application.

mkdir build&&cd buildcmake ..&& make -j8

If Flashlight or ArrayFire are installed in nonstandard paths via a customCMAKE_INSTALL_PREFIX, they can be found by passing

-Dflashlight_DIR=[PREFIX]/usr/share/flashlight/cmake/ -DArrayFire_DIR=[PREFIX]/usr/share/ArrayFire/cmake

when runningcmake.

• Facebook page:https://www.facebook.com/groups/717232008481207/
• Google group:https://groups.google.com/forum/#!forum/wav2letter-users
• Contact:vineelkpratap@fb.com,awni@fb.com,qiantong@fb.com,jacobkahn@fb.com,antares@fb.com,avidov@fb.com,gab@fb.com,vitaliy888@fb.com,locronan@fb.com

wav2letter++ is MIT-licensed, as found in theLICENSE file.