Adding Personality to Neutral Speech Synthesis Voices
A synthetic voice personifies the system using it. Previous work has shown that using sub-corpora with different voice qualities (e.g. tense and lax) can be used to modify the perceived personality of a voice as well as adding expressive and emotional functionality. In this work we explore the use of LPC source/filter decomposition together with modification of the residual to artificially add voice quality sub-corpora to a voice without recording bespoke data. We evaluate this artificially enhanced voice against a baseline unit selection voice with pre-recorded sub-corpora. Although artificial modification impacts naturalness, it has the advantage of adding emotional range to voices where none was recorded in the source data, deals with data sparsity issues caused by sub-corpora, and results in significant effects in terms of perceived emotion.
KeywordsVoice modification Glottal signal modelling Glottal vocoding Speech synthesis Unit selection Expressive speech synthesis Emotion Prosody Artificial personality
This work was supported by the European Union’s Horizon 2020 research and innovation programme under grant agreement No. 645378 (Aria VALUSPA).
- 1.Schröder, M.: Emotional speech synthesis: A review. In: Seventh European Conference on Speech Communication and Technology (2001)Google Scholar
- 3.Aylett, M.P., Vinciarelli, A., Wester, M.: Speech synthesis for the generation of artificial personality. IEEE Trans. Affect. Comput. (2017)Google Scholar
- 5.Aylett, M., Pidcock, C.: Adding and controlling emotion in synthesised speech. Pat no. GB2447263A, September 2008Google Scholar
- 6.Gibiansky, A., et al.: Deep voice 2: multi-speaker neural text-to-speech. In: Advances in Neural Information Processing Systems, pp. 2966–2974 (2017)Google Scholar
- 8.Huber, S., Roebel, A.: On glottal source shape parameter transformation using a novel deterministic and stochastic speech analysis and synthesis system. In: Interspeech 2015 (2015)Google Scholar
- 9.Shechtman, A., Shechtman, S., Rendel, A.: Semi parametric concatenative TTS with instant voice modification capabilities. In: Interspeech 2017 (2017)Google Scholar
- 10.Drugman, T., Wilfart, G., Dutoit, T.: A deterministic plus stochastic model of the residual signal for improved parametric speech synthesis. In: Interspeech 2009 (2009)Google Scholar
- 12.Csap, T.G., Nmeth, G., Cernak, M., Garner, P.N.: Modeling unvoiced sounds in statistical parametric speech synthesis with a continuous vocoder. In: 24th European Signal Processing Conference (EUSIPCO), pp. 184–194 (2016)Google Scholar
- 15.Fant, G., Liljencrants, J., Lin, Q.G.: A four-parameter model of glottal flow. STL-QPSR 26(4), 001–013 (1985)Google Scholar
- 17.Fant, G.: The LF-model revisited. Transformations and frequency domain analysis. Speech Trans. Lab. Q. Rep. R. Inst. Tech. Stockh. 2(3), 40 (1995)Google Scholar
- 18.Brookes, M.: VOICEBOX: speech processing toolbox for MATLAB. http://www.ee.ic.ac.uk/hp/staff/dmb/voicebox/voicebox.html. Accessed 13 Oct 2017
- 19.Kominek, J., Black, A.W.: The CMU arctic speech databases. In: Fifth ISCA Workshop on Speech Synthesis (2004)Google Scholar