ASR-events-OC17

Introduction

The modern society demonstrates clear mutual influence among languages, e.g., Mandarin to minor languages in China, and English to other languages in the world. This leads to a clear mixlingual phenomenon, i.e., some words of a foreign (or target, embedded) language are embedded in a host (or source, matrix) language. This mixlingual phenomenon results in a serious problem in speech recognition (ASR).

Based on the success of the first MixASR-CHEN 2016 challenge, the MixASR-CHEN 2017 challenge follows the same theme of Mixlingual ASR. The task is more challenging in several ways:

• There are more utterances that involve multiple English words in the test data
• There are more English words that are not in the CMU dictionary
• There are more English phrases that involve multiple English words

Challenge details

• The database information is here.
• The challenge plan is here.
• The tools that can be used are here.
• The Kaldi baseline is here.
• Registration and result submission here.
• Participants from both academy and industry are welcome.
• As the first challenge, we will base the challenge on the OCOCOSDA forum and will release the results on OCOCOSDA 2017. Challenge participants are highly recommended to submit their system as a paper to the conference.

Important date

• May 1st: training/dev dataset release
• July 15: OC16-CE80 test data release
• July 18: OC16-CE80 test data release
• July 29: Paper submission deadline
• OC2017: challenge result release

Organizers

• Dong Wang (Tsinghua University)
• Zhiyuan Tang (Tsinghua University)
• Qing Chen (Speech Ocean), chenqing@speechocean.com