The SIMPLE4ALL project is creating speech synthesis technology that learns from data with little or no expert supervision and continually improves itself, simply by being used.
Publications
Public deliverables
- D1.6 - to be published after final review
- D2.1: Initial synthesis front end description
- D2.2: Description of the final version of the front-end
- D3.1: Automatic parameterization with WLP-based GIF-techniques
- D3.2: Report describing initial version of deep layered models
- D3.3: Evaluation of the new speech signal model in conjunction with HMM-based acoustic models
- D3.4: Report describing the final version of the deep layered models
- D3.5: Final evaluation report
- D4.1: Learning from user-provided data
- D4.2: Online learning for improving one, or more than one component
- D4.3 – to be published after final review
- D5.1: Acoustic and prosodic analysis of genres and speaking styles
- D5.2: Final report on the implementation and evaluation of genre classification
- D6.1: Project website / remote collaboration tools
- D6.3 - to be published after final review
- D6.4 - to be published after final review
- D6.7 - to be published after final review
Internship reports
- Bajibabu Bollepalli: Effect of MPEG audio compression on HMM-based speech synthesis
- Jiunn Lin Wong: Realisation and simulation of the Mel log spectrum approximation filter
- Dalia Popescu: Unsupervised Text Syllabification with Information from Audio
- Srikanth Ronanki: Syllable based models for prosody modelling in HMM based speech synthesis
- Jose Moreno: Effects of Noise on a Speaker-Adaptive Statistical Speech Synthesis System
- Arturo Romero Blanco : Spanish Emotional Speech Synthesis
- Jaime Lorenzo-Trueba: Towards Cross-lingual Emotion Transplantation
- Ascension Gallardo: A Comparison of Open-Source Segmentation Architectures
Academic publications
Acronyms Adaptation affective agent audiobook data Contextual Factors Deep neural network discriminative training DNN expressive speech synthesis formants glottal flow glottal inverse filtering glottal source GlottHMM HMM intelligibility Lombard speech machine learning morfessor robust spectrum estimation speaker diarization speaking styles speech alignment speech analysis Speech synthesis statistical parametric speech synthesis transplantation unsupervised learning vector space model weighted linear prediction
| (2014): DNN-based stochastic postfilter for HMM-based speech synthesis. In: Proc. Interspeech 2014, pp. 1954-1958, Singapore, 2014. |
| (2014): Morfessor FlatCat: An HMM-Based Method for Unsupervised and Semi-Supervised Learning of Morphology. In: Proceedings of COLING 2014, the 25th International Conference on Computational Linguistics, pp. 1177–1185, Dublin City University and Association for Computational Linguistics, Dublin, Ireland, 2014. |
| (2013): Effect of MPEG Audio Compression on HMM-based Speech Synthesis. In: Proc. Interspeech 2013, 2013. |
| (2013): A Comparative Evaluation of Vocoding Techniques for HMM-based Laughter Synthesis. In: Proc. ICASSP 2013, 2013. |
