SHACHI - Language Resource Metadata Database

The RIKEN Spoken Dialogue Corpus consists of simulated dialogues (word processor specialist - the user), lasting up to two hours at longest, and monologues (usage guidance by word processor specialists). The corpus comes with corresponding morphologically tagged transcripts.

The corpus contains speech recordings and their orthographic transcriptions, with Kana or Romaji representing pronunciation of sentences, of 306 speakers reading excerpts from the Mainichi Newspaper and the ATR 503 Phonetically Balanced Sentences. All utterances and sentences are in the Japanese language.

The S-JNAS is a collection of read speech data of Japanese elderly people reading newspaper article sentences and ATR phonemically balanced sentences from the read-out text of the JNAS (Japanese Newspaper Article Sentence).

The ATR Chinese Hotel Reservation Dialogue is the Chinese parallel speech corpus included in the multi-lingual project, which has collected multi-lingual hotel reservation dialogues, consisting of Japanese, English, and Chinese speeches.

12 volumes of manually tagged (lemma, POS, syntax) Chinese language text used in the Singapore schools, covering levels from Primary 1 to Primary 6.

The CSTC-Flight is a domain-specific corpus containing flight enquiry and reservation domain telephone conversations, taken from the real life. All the dialogues are fully spontaneous since the coustomers are not aware of being monitored and recorded. Corresponding transcripts are included in the corpus for only about half of the speech data.

CUCorpora is a large scale Cantonese spoken language corpora, made up of several sub-corpora designed for different specific domain of applications. The sub-corpoora are; (1) 1800 Cantonese syllables with pitch-marking (CUSYL), (2) Multi-syllabe short phrases covering most common Cantonese syllables (CUWORD), (3) Phoneticaly-rich read Cantonese sentences (CUSENT), (4) Cantonese digit strings of length from single digit to 16 digit (CUDIGIT), and (5) Cantonese command words simulating a navigation control scenario (CUCMD). All sub-corpora are manually transcribed.

CUSYL is a part of CUCopora, a large scale Cantonese spoken language corpora. It is a collection of 1800 Cantonese syllables with pitch-marking, covering all valid syllables as well as common lazy and colloquial pronunciations. The corpus also includes manual transcripts. The package comes with CUWORD (Version 1.0).

CUWORD is a part of CUCopora, a large scale Cantonese spoken language corpora. It is a collection of 2500 multi-syllabe short phrases covering most common Cantonese syllables. The corpus also includes manually verified phonemic transcription provided for each utterance. The package comes with CUSYL. (Version 1.0).

CUSENT is a part of CUCopora, a large scale Cantonese spoken language corpora. It is a large collection of spoken Cantonese sentences designed to be phonetically rich. The corpus also includes manually verified phonemic transcription.