Abstract
Text-dependent speaker verification over degraded radio channel is a challenging task. To better understand the research problem, the Institute for Infocomm Research (I2R) of Singapore has collected a corpus of voice recordings transmitted over marine VHF. Built as an extension of the RSR2015 database, the VHF-RSR2015 consists of recordings from 300 speakers of Part I of the RSR2015 database transmitted over VHF channel. Extending the RSR2015 database, we would like to facilitate the study of the VHF channel effect, therefore, keeping the acoustic environment the same as that of RSR2015. Performance benchmark of a text-dependent speaker verification engine is given as reference on the original RSR2015 database recorded at a sampling frequency of 16kHz, on a sub-sampled version of the same dataset at 8kHz and on the data after transmission through the VHF channel.
| Original language | English |
|---|---|
| Pages (from-to) | 1322-1326 |
| Number of pages | 5 |
| Journal | Proceedings of the Annual Conference of the International Speech Communication Association, INTERSPEECH |
| Publication status | Published - Sept 2014 |
| Externally published | Yes |
| Event | 15th Annual Conference of the International Speech Communication Association: Celebrating the Diversity of Spoken Languages, INTERSPEECH 2014 - Singapore, Singapore Duration: 14 Sept 2014 → 18 Sept 2014 |
Keywords
- Database
- Speaker verification
- Text-dependent
- VHF
ASJC Scopus subject areas
- Language and Linguistics
- Human-Computer Interaction
- Signal Processing
- Software
- Modelling and Simulation