To commemorate the Homecoming Day, the abstracts of the dissertations of the members who attended at the time were used to compose special songs, using the Suno service.
Lossless versions of each album, with liner notes can be downloaded from:
01 Reunion Album
02 The Philosophers
03 LeemDongZ Saga
04 Academic Rave
All copyrights belong to the original authors of the lyrics, i.e., the individual dissertations
1. 2024/06/10 Homecoming Day Commemorative Album
Each song was composed using the abstracts of the dissertations of the members who gathered at that time. Tracks 1-11 are sorted by phonetical order of author names. Track 12 is a remix version of Track 10, and Tracks 13 through 15 are bonus tracks that compile the theses of three people who researched the common theme of pipelining in chronological order.
| 01 | 고봉구, Cycle-consistency를 이용한 다수 화자 한국어 음성변환 | |
| 02 | 양수원, 음성 변환을 위한 cycleVAE의 성능 향상 연구 | |
| 03 | 오현우, 지연 보상을 적용한 파이프라인 방식 확률적 경사 하강법 | |
| 04 | 이건녕, 다수 화자 음성 변환을 위한 VAE의 잠재 공간 disentanglement와 변환 경로 학습 | |
| 05 | 이효원, Pipelined stochastic gradient descent for training deep neural network models | |
| 06 | 임단, Improving seq2seq by revising attention mechanism for speech recognition | |
| 07 | 임성균, 심층신경망 기반 음성 인터페이스의 환경 적응을 위한 선형 스펙트럼 특징 변환 | |
| 08 | 장봉원, 확률적 경사 하강법의 파이프라인 병렬화에서 지연 기울기 완화 기법 연구 | |
| 09 | 장형필, 음성 변환 기술을 활용한 Zero-shot 화자 익명화 연구 | |
| 10 | 정동화, Triple attention을 이용한 음성 인식 | |
| 11 | 정창현, 적대적 생성 신경망 기반 보코더를 이용한 음성 변환 | |
| 12 | (Bonus) 정동화, Triple attention을 이용한 음성 인식 | |
| 13 | (Bonus) 이효원, Pipelined stochastic gradient descent for training deep neural network models | |
| 14 | (Bonus) 오현우, 지연 보상을 적용한 파이프라인 방식 확률적 경사 하강법 | |
| 15 | (Bonus) 장봉원, 확률적 경사 하강법의 파이프라인 병렬화에서 지연 기울기 완화 기법 연구 |
2. Doctoral Dissertations
This song was composed using the abstracts of the doctoral dissertations of all the past Ph.Ds in the lab. Track 0, of course, is the doctoral dissertation of the advisor
| 00 | D. Yook, Robust Speech Recognition Using Neural Networks and Hidden Markov Models -Adaptations Using Non-linear Transformations- | |
| 01 | D. Kim, Acoustic model adaptation using linear spectral transformation for robust speech recognition | |
| 02 | Y. Cho, Robust speaker localization using steered response voice power | |
| 03 | H. Lee, Multi-channel voice activity detection using multi-source | |
| 04 | I. C. Yoo, Robust voice activity detection using formant frequencies | |
| 05 | H. Lim, Speaker selective source localization for non-trivial noise environments | |
| 06 | T. Lee, Parallel SRP-PHAT for GPUs |
3. Seonggyun Lim’s Doctoral Degree Commemorative Single Album
This song was specially produced to commemorate the doctoral degree of Seonggyun Lim, who completed his master’s degree at our lab and his doctoral degree abroad. In keeping with the theme of his “Speech Emotion Recognition System,” composing process aimed for a song filled with strong emotions that challenge the limits of AI composition
| 01 | S. G. Leem, Speech Emotion Recognition in the Presence of Background Noise |
4. Homecoming Day 1st Anniversary Album
To celebrate the irony of a commemoration of the commemoration, we have recreated the 1st anniversary album using the original lyrics and a unified musical style. This time, the songs are sorted in reverse phonetical order of author names. Apollonian (logical abstracts) meets Dionysian (maniac passion), Nietzsche will be proud!