ML-SUPERB: Multilingual Speech Universal PERformance Benchmark.
Jiatong ShiDan BerrebbiWilliam ChenHo-Lam ChungEn-Pei HuWei-Ping HuangXuankai ChangShang-Wen LiAbdelrahman MohamedHung-yi LeeShinji WatanabePublished in: CoRR (2023)
Keyphrases
- multi lingual
- maximum likelihood
- speech recognition
- digital libraries
- cross lingual
- speech synthesis
- real world
- spoken language
- language resources
- broadcast news
- audio visual
- comparative analysis
- language independent
- spontaneous speech
- vocal tract
- recognition engine
- data sets
- multimodal interfaces
- text to speech synthesis
- log analysis
- speaker recognition
- automatic speech recognition
- speech signal
- n gram
- pattern recognition
- multimedia