Overview:
Speech is the most natural, powerful and universal human-machine/computer interface. Therefore our focus is to make speech recognition more robust for our daily life. To demonstrate the concept and our research results, a spoken dialog-based multimedia retrieval prototype will be built to integrate both voice and data services under internet environment.Research Direction:
- Robust speech recognition
- Robust speaker and language recognition
- Handset and channel effect compensation
- Application of prosodic information
- Computer assisted language learning (CALL)
- Text-to-speech (TTS)
- Blizzard Challenge
- Robust spoken dialogue system
- Distribution Speech Recognition (DSR)
Annual Speech Processing Workshop:
Group Project Link:
Demonstration:
Projects
- Blizzard challenge 2009
(Text-to-Speech)
-
50 samples of our submission for Mandarin Hub Task:
NTUT_Blizzard2009_MH_sample.tgz (wav, ~6M, Mandarin)
NTUT_Blizzard2009_MH_STRAIGHT_sample.tgz (wav, ~6M, Mandarin)
-
- ITS Spoken Dialogue System
- Year 2004: In-vehicle speaking assistant system (spoken dialog system for GPS car navigation around HsinChu Science Park and its neighborhood)
- Year 2005: Enhanced in-vehicle speaking assistant system
-
Microphone array
-
Robust parser and spontaneous speech recognition
-
Corpus-based TTS
-
Undergraduate Projects
- Speech recognition-enabled
automatic telephone directory assistant system for National Taipei University
of Technology (NTUT's 104 directory assistant service )
NTUT's extension number 5104 = NTUT's 104 directory assistant service ( +886-2-27712171 ext. 5104)
Example: You want to find committee Cai-Jia Chun " 電子系主任賴柏洲" (dian zi xi zhu ren lai bo zhou)，you can use the following four methods
- Example 1: call extension 5104，speak ”賴柏洲 ” (lai bo zhou)
- Example 2: call extension 5104，speak ”電子系賴柏洲 ” (dian zi xi lai bo zhou)
- Example 3: call extension 5104，speak ”電子系主任 ” (dian zi xi xhu ren)
- Example 4: call extension 5104，speak ”電子系 ” (dian zi xi)
- (Phone book.txt) and
(Phone_book.xls)
(the first column: names or department, the second column:
pronunciations, the third column: extensions of telephone system).
(Update: 2006/01/26)
corresponding person :Zhi-Ren Zeng , email : s9360382@yahoo.com.tw extension:2247, cell-phone:0919550114
- Communication Platform for Distributed Speech Recognition(DSR) VoIP. Text Chat & Face Simulation
Windows video
(wmv, 20.1M, Mandarin)
Prof. Sagayama, Tokyo Univ.
visit our demo
- Speaker Verification Security System
Prof. Fujisaki, Tokyo Univ. visit
our demo
- Multi modal spoken dialog system-based multimedia retrieval system
Windows
video (wmv, 15.8M, Mandarin)
- Voice over IP (VOIP) system and real-time ETSI extended advanced DSR front-end codec (ETSI ES 208 212)
Windows video
(wmv, 20.4M, Mandarin)
Communication DSP Lab.
- OFDM Transmitter and Receiver Implementation Using DSK6416
- DSP-based OFDM 語音即時通訊系統
以分散式語音辨認為基礎之網路電話語音自動總機
Latest Poster&Presentation:
A Reference Model Weighting-based Method for Robust Speech Recognition
Multimedia Network Communicate Platform
Latent Prosody Analysis for Robust Speaker Identification
Test Norm-Based Speaker Verification Security System
Lab. Overview:
-
Speech Processing (Speech)
- Overview：
- Cooperation with NCTU speech lab., professor Sin-Horng, Chen and professor Yi-Ru, Wang
- National Science Council (NSC) 3 years project (2004.8~2007.7)"Robust speaker verification for spoken dialogue system "
- NCTU's ITS project, spoken dialogue system using galaxy communicator
- Graduate student: four
- Undergraduate: student: seven
- Workstation: seven
- Dialog Card: two
- Pictures:
Linux workstation Lab Meeting Room
-
Teaching:
- Digital signal processing
- Digital signal processing laboratory
- TI DSK6711
- Digital signal processing laboratory
- Speech processing
- Random processing
- Communication signal processing
- Communication signal processing lab.
- TI DSK6416 + Signalware AED-101 80M Wideband AD/DA
- Communication signal processing lab.
- Digital signal processing
-
Communication/Digital Signal Processing (CommDSP/DSP Lab.)
- Overview：
- Dsk6711：twenty
- Dsk6416：thirty
- Evm6701：two
- Course
- Communication signal processing laboratory (graduate)
- Digital signal processing laboratory (undergraduate)
- Communication Signal Processing laboratory
- Overview of TI DSK6416
code composer studio, DSP/Bios- Real-time/Embedded DSP framework
- Vocoder (GSM6.10)
- Convolution code and Viterbi decoding
- QAM baseband Transmitter/Receiver, Timing synchronization, vocoder
- Channel Model
AWGN, multipath fading channel- OFDM baseband Transmitter/Receiver, FFT, Frequency/Phase synchronization
Channel Estimation, Scrambler/interleaver, Convolution code/Viterbi decoder
- Digital signal processing laboratory
- Overview of TI DSK6711
- AM modulation and demodulation
- EMIF and memory
- Timer and interrupt service
- DMA
- Calling assembly language function from C language (optional)
- Sound recording and playing
- Data transfer between PC and DSK (optional)
- FFT
- Embedded system
- Karaoke effect (changing voice, echo)
- Digital filter (equalizer)
- Music synthesis
Server:
Contact Information
- Tel.
- +886-2-2771-2171 ext. 2247
- Fax
- +886-2-2731-7120
- Address
- 1, Sec. 3, ChungHsiao E. Rd. Taipei, Taiwan (National Taipei University of Technology composite accommodation 406)
