CHiME 2016, 13th Sep
San Francisco, US
- Emmanuel Vincent
- Shinji Watanabe MERL, USA
- Jon Barker &
Univ. of Sheffield, UK
- More info
- Registration: Jun 10th
- Submission: Aug 19th
- Notification: Aug 29th
- Workshop: Sep 13th
- More info
Speech Processing in Everyday Environments (CHiME 2016)
13th September, Google, San Francisco (a satellite event of Interspeech 2016)
The workshop took place in September 2016. This website remains as a record of the event including the programme of talks, video of most of the presentations and access to the full proceedings.
This one-day workshop will bring together researchers from the fields of computational hearing, speech enhancement, acoustic modelling and machine learning to discuss the robustness of speech processing in everyday environments, i.e., real-world conditions with acoustic clutter, where the number and nature of the sound sources is unknown and changing over time.
Relevant research topics include (but are not limited to)
- training schemes: data augmentation, semi-supervised training,
- speaker localization and beamforming,
- single- or multi-microphone enhancement and separation,
- robust features and feature transforms,
- robust acoustic and language modeling,
- robust speech recognition,
- robust speaker and language recognition,
- robust paralinguistics,
- cross-environment or cross-dataset performance analysis,
- environmental background noise modelling.
Papers reporting evaluation results on the CHiME-4 dataset or on other datasets are both welcome.
The CHiME-4 Challenge
As a focus for discussion, the workshop will host the CHiME-4 Speech Separation and Recognition Challenge. Following the success of the CHiME-3 Challenge which attracted 25 international teams, CHiME-4 revisits the CHiME-3 data, i.e., utterances recorded via a 6-microphone tablet device in challenging noisy environments. The difficulty is increased by reducing the number of microphones. To find out more please visit the Challenge Website