Speech synthesis technology is used in computer system for memorizing people’s actual pronunciation, reorganizing the collected information and building-up continuous word flow. Speech synthesis technology applies in many fields, for instance, economic, livelihood and production, technology and so on. A high quality speech corpus is no doubt the basis of a good speech synthesis system. The speech synthesis corpus we have made has advantages of extensive coverage and high consistency, etc.

1. High quality

1) Professional speakers: the speakers should have adaptability training and other specific training for making speech synthesis corpus; it is necessary to have strict selection of those speakers.

2) Recording environment: professional recording studio with professional recording equipments. Variation of parameters in the recording studio is within a certain small range. Try to keep all the conditions as consistent as possible during the speech recording period. The speech data has good signal-to-noise (SNR) rate after sampling.

3) Reading style of speakers: mostly news-broadcast reading style is required, but not limit to it. We can adjust reading styles according to requirements of different clients. The whole recording is strictly monitored to make sure the correctness and consistence of all speech data.

4) Post-process of the speech data: the high signal-to-noise rate is well preserved. We are very restricting on speech labeling to keep good consistence among all labelers.

2. High efficient

1) Procedure of speech recording: To make the recording procedure smooth and efficient, the reading texts will be pre-corrected; monitoring the speakers in the whole process and ask the speakers to get familiar with the reading text before recording. To ensure the high quality, we will finish the speech corpus in a very short period with proper arrangement of recording schedule and space between reread items.

2) Post-processing of the speech corpus: develop and modify necessary processing tools; make flexible adjustment according to certain demands; in order to achieve a high coherence and fast speed in development, we will have professional orientation training; build up strict Quality monitoring system.

●Our services:
1) Speech synthesis corpus can be consigned to us including the following aspects: collection of reading material, speech data recording, speech data pack-up, annotation of speech data etc. The clients can choose one aspect or several as a group.

2) According to different requirement of our clients, the collection of speech material can be selected from different speech sources of different fields. Till now, we have speech corpus of Mandarin Chinese, some different dialects of Chinese, English and Spanish. New languages are also taken into consideration.

3) Contents of the speech corpus: the abuse of syllable, neutral tone, digit strings, English letters, Greece letters, Chinese single sentences, Overlapping words, common used English words and so on. Usually the corpus composed of syllables, phrases, sentences and discourses, etc.

4) Speech corpus annotation includes boundaries of syllable, phoneme, metrical levels, lexical pronunciation, actual pronunciation, words in lexicons, and metrical terms etc.

Rely on our professional experience for years, we guaranteed to provide reasonable product schedule for speech synthesis corpus.