Acoustic Durational Properties of Sonorant as Syllable Boundaries in Text-to-Speech Synthesis
Understanding the acoustic properties of phonetic unit is the key for concatenating natural speech in text-to-speech synthesis. In this paper, two sonorant, /n/ and /l/, are chosen as the target phonemes to learn their acoustic durational behaviors in CV and VC phonetic sequences. The results show that both share similarities and differences in durational behaviors with two variables of stress and position. Moreover, /n/ behaves more stable than /l/ both as onsets and codas. Their segmental application in TTS system (Text-to-Speech system) is discussed finally.
Keywords/N/ /L/ CV VC Duration Stress Position Boundary Text-to-speech synthesis
- 2.Nguyen N, Hawkins S (1999) Implications for word recognition of phonetic dependencies between syllable onsets and codas. In: Ohala JJ, Hasegawa Y, Ohala M, Granville D, Bailey AC (eds) Proceedings of the XIVth international congress of phonetic sciences. University of California, Berkeley, vol 1(3), pp 647–650Google Scholar
- 6.van Son RJJH (2001) Strong interaction between factors influencing consonant duration. Foreign Lang Teach Res 7(2):12–19 Google Scholar