Embedded Files
Note: This site had been move to https://sites.google.com/nycu.edu.tw/speechlabx
Formosa Speech Recognition Challenge 2018 (FSR-2018) is the first event of the Formosa Speech in the Wild (FSW) project, which is an official ISCSLP-2018 activity organized by Taipei University of Technology (NTUT).
Formosa Speech Recognition Challenge 2018 (FSR-2018) is the first event of the Formosa Speech in the Wild (FSW) project, which is an official ISCSLP-2018 activity organized by Taipei University of Technology (NTUT).
Since the language habits of Taiwanese people are different from other Mandarin speakers (both accents and cultures). The current general-purpose automatic Mandarin speech recognition solutions are not suitable for supporting Taiwanese people’s daily life. In order to boost the development of Taiwan-specific speech recognition techniques. We are now calling for and welcome participants from both academic and industrial sectors to FSR-2018. Especially, students are encouraged to opt-in of being considered for the Student Award.
Since the language habits of Taiwanese people are different from other Mandarin speakers (both accents and cultures). The current general-purpose automatic Mandarin speech recognition solutions are not suitable for supporting Taiwanese people’s daily life. In order to boost the development of Taiwan-specific speech recognition techniques. We are now calling for and welcome participants from both academic and industrial sectors to FSR-2018. Especially, students are encouraged to opt-in of being considered for the Student Award.
Call-for-Participants
Call-for-Participants
Please help us to distribute this Call-for-Participants.
Draft version of the challenge description paper for your reference: "Formosa Speech Recognition Challenge 2018- Data, Plan and Baselines"
Key Messages
Key Messages
Real-Life/Multi-Gene/Spontaneous Taiwanese Mandarin Speech from broadcast radio show series
Unique language habits in Taiwan (different accents and cultures)
Task
Task
Build an automatic speech recognizer (ASR) that is suitable for recognizing Spontaneous Taiwanese Mandarin Speech (broadcast radio).
Database
Database
This challenge is based on the "NER-Trs-Vol1" corpus selected from National Education Radio Archive.
"NER-Trs-Vol1" is manually transcribed and consists 11 radio show series, in total about 126.6 hours (training set).
This data is released under a license for Non-Commercial Use Only. Please read and accept the License.
Here is some Audio/Transcription Samples for your reference.
Baseline Scripts: Kaldi-based baseline recipes are provided in Github for new researchers/students to develop their systems easily and quickly (kaldi-asr:master pull request #2474).
Result/Award Announcement
Result/Award Announcement
Result/Award will be announced in ISCSLP 2018 (http://iscslp2018.org), Taipei, 26 Nov. 2018
Important Dates
Important Dates
2018/06/01 --- Registration Open, Training Data Release
2018/08/01 --- Pilot-Test (dry-run only) Data Release
2018/08/08 --- Pilot-Test (dry-run only) Result Submission
2018/08/15 --- Pilot-Test (dry-run only) Performance Notification
2018/10/31 --- Registration Close
2018/11/01 --- Final-Test Data Release
2018/11/08 --- Final-Test Result Submission
2018/11/15 --- Final-Test Performance Notification (released)
2018/11/23 --- Paper Submission (deadline extended)
2018/11/26 --- Result/Award Announcement, Satellite Workshop of ISCSLP 2018
PS: Pilot-Test (dry-run) is only used to make sure everything for final-test is fine, not for scoring!!!
Contact
Contact
Yuan-Fu Liao (廖元甫)
Associate Professor, Department of electronic Engineering, National Taipei University of Technology
Sponsors
Sponsors
Report abuse