Resources for Automatic Speech Recognition

The WaveSurfer Automatic Speech Recognition Plugin

The plugin is described in the following paper:

• You can find the source code and installation instructions at https://github.com/giampierosalvi/ASRWavesurferPlugin
• The latest release is at https://github.com/giampierosalvi/ASRWavesurferPlugin/releases
• Language pack for Swedish
• Language pack for English

HTK based WaveSurfer Automatic Speech Recognition Plugin

This is an earlier version of the plugin above. The difference is that this version is based on HVite (HTK). This plugin is still used for educational purposes in the DT2112 Speech Technology course at KTH.

• hvite.plug
• To install, run wavesurfer once to create the configuration directory, and then copy the file into ~/.wavesurfer/1.8/plugins/
• No language and acoustic models are defined, so the plugin is relatively useless outside the scope of the course.
• Note: you will need a working version of HTK installed on your system for this to work.

ASR models for Swedish (Version 0.1)

This is a collection of acoustic and language models that were used in the paper:

• package: asr-models-swedish-0.1.tgz (2.1GiB, MD5: a9ddfce26772ee5590f3650a1b2341ae)
• language: Swedsih
• format: HTK, Julius
• sampling rates: 8/16kHz
• features: MFCC/PLP
• acknowledgements: this research has been supported by CLT (Gothenburg Center for Language Technology)
• Test sets definition for the LREC 2014 experiments: testlistsLREC2014.tgz (109KiB, MD5: 2fa941ef10a4d1d36f3fb015ec2110b7)

This work is licensed under a Creative Commons Attribution 3.0 Unported License.