The UPC TTS System Description for the 2008 Blizzard Challenge

This paper presents the UPC TTS system named Ogmios. It was used to generate the voices in UK English and Mandarin Chinese for Blizzard Challenge 2008. Ogmios is a system based on unit-selection using acoustic and phonetic features both in target and concatenation costs. Most of the modules of Ogmios rely on data driven techniques. This evaluation confirms that this framework allows fast development of synthetic voices in new languages that were not previously covered by the TTS: Mandarin Chinese. The work in the TTS was devoted to prepare Ogmios to synthesize Mandarin Chinese, taking into account some particularities of the language regarding its writing and intonation. In the case of UK English we put the focus on rhythm, pauses and phonetic transcription. The evaluators scored the UK English voices in an average of 3. High WER indicates intelligibility problems. Mandarin Chinese voice was scored with a lower score, near 2. Several measures indicate intelligibility and quality problems in this synthetic voice.