SHACHI - Language Resource Metadata Database

Desktop/Microphone
The GlobalPhone corpus developed in collaboration with the Karlsruhe Institute of Technology (KIT) was designed to provide read speech data for the development and evaluation of large continuous speech recognition systems in the most widespread languages of the world, and to provide a uniform, multilingual speech and text database for language independent and language adaptive speech recognition as well as for language identification tasks.

The entire GlobalPhone corpus enables the acquisition of acoustic-phonetic knowledge of the following 20 spoken languages: Arabic (ELRA-S0192), Bulgarian (ELRA-S0319), Chinese-Mandarin (ELRA-S0193), Chinese-Shanghai (ELRA-S0194), Croatian (ELRA-S0195), Czech (ELRA-S0196), French (ELRA-S0197), German (ELRA-S0198), Hausa (ELRA-S0347), Japanese (ELRA-S0199), Korean (ELRA-S0200), Polish (ELRA-S0320), Portuguese (Brazilian) (ELRA-S0201), Russian (ELRA-S0202), Spanish (Latin America) (ELRA-S0203), Swedish (ELRA-S0204), Tamil (ELRA-S0205), Thai (ELRA-S0321), Turkish (ELRA-S0206), Vietnamese (ELRA-S0322).

In each language about 100 sentences were read from each of the 100 speakers. The read texts were selected from national newspapers available via Internet to provide a large vocabulary (up to 65,000 words). The read articles cover national and international political news as well as economic news. The speech is available in 16bit, 16kHz mono quality, recorded with a close-speaking microphone (Sennheiser 440-6) and same recording equipment for all languages. The transcriptions are internally validated and supplemented by special markers for spontaneous effects like stuttering, false starts, and non-verbal effects like laughing and hesitations. Speaker information like age, gender, occupation, etc. as well as information about the recording setup complement the database. The entire GlobalPhone corpus contains over 450 hours of speech spoken by more than 1900 native adult speakers.

Data is shortened by means of the shorten program written by Tony Robinson, available from Softsound's web page: http://www.softsound.com/ linux distributions, or simulated versions such as cygwin. Alternatively, the data could be delivered unshorten.

The Swedish corpus was produced using the Goeteborgs-Posten newspaper. It contains recordings of 98 speakers (50 males, 48 females) recorded in Stockholm and Vaernamo, Sweden. The following age distribution has been obtained: 9 speakers are below 19, 50 speakers are between 20 and 29, 12 speakers are between 30 and 39, 11 speakers are between 40 and 49, and 16 speakers are over 50.

Desktop/Microphone
The GlobalPhone corpus developed in collaboration with the Karlsruhe Institute of Technology (KIT) was designed to provide read speech data for the development and evaluation of large continuous speech recognition systems in the most widespread languages of the world, and to provide a uniform, multilingual speech and text database for language independent and language adaptive speech recognition as well as for language identification tasks.

The entire GlobalPhone corpus enables the acquisition of acoustic-phonetic knowledge of the following 20 spoken languages: Arabic (ELRA-S0192), Bulgarian (ELRA-S0319), Chinese-Mandarin (ELRA-S0193), Chinese-Shanghai (ELRA-S0194), Croatian (ELRA-S0195), Czech (ELRA-S0196), French (ELRA-S0197), German (ELRA-S0198), Hausa (ELRA-S0347), Japanese (ELRA-S0199), Korean (ELRA-S0200), Polish (ELRA-S0320), Portuguese (Brazilian) (ELRA-S0201), Russian (ELRA-S0202), Spanish (Latin America) (ELRA-S0203), Swedish (ELRA-S0204), Tamil (ELRA-S0205), Thai (ELRA-S0321), Turkish (ELRA-S0206), Vietnamese (ELRA-S0322).

In each language about 100 sentences were read from each of the 100 speakers. The read texts were selected from national newspapers available via Internet to provide a large vocabulary (up to 65,000 words). The read articles cover national and international political news as well as economic news. The speech is available in 16bit, 16kHz mono quality, recorded with a close-speaking microphone (Sennheiser 440-6) and same recording equipment for all languages. The transcriptions are internally validated and supplemented by special markers for spontaneous effects like stuttering, false starts, and non-verbal effects like laughing and hesitations. Speaker information like age, gender, occupation, etc. as well as information about the recording setup complement the database. The entire GlobalPhone corpus contains over 450 hours of speech spoken by more than 1900 native adult speakers.

Data is shortened by means of the shorten program written by Tony Robinson, available from Softsound's web page: http://www.softsound.com/ linux distributions, or simulated versions such as cygwin. Alternatively, the data could be delivered unshorten.

The Tamil corpus was produced using the Thinaboomi Tamil Daily newspaper. It contains recordings of 47 speakers (gender unspecified) recorded in India. No age distribution is available.

Desktop/Microphone
The GlobalPhone corpus developed in collaboration with the Karlsruhe Institute of Technology (KIT) was designed to provide read speech data for the development and evaluation of large continuous speech recognition systems in the most widespread languages of the world, and to provide a uniform, multilingual speech and text database for language independent and language adaptive speech recognition as well as for language identification tasks.

The entire GlobalPhone corpus enables the acquisition of acoustic-phonetic knowledge of the following 20 spoken languages: Arabic (ELRA-S0192), Bulgarian (ELRA-S0319), Chinese-Mandarin (ELRA-S0193), Chinese-Shanghai (ELRA-S0194), Croatian (ELRA-S0195), Czech (ELRA-S0196), French (ELRA-S0197), German (ELRA-S0198), Hausa (ELRA-S0347), Japanese (ELRA-S0199), Korean (ELRA-S0200), Polish (ELRA-S0320), Portuguese (Brazilian) (ELRA-S0201), Russian (ELRA-S0202), Spanish (Latin America) (ELRA-S0203), Swedish (ELRA-S0204), Tamil (ELRA-S0205), Thai (ELRA-S0321), Turkish (ELRA-S0206), Vietnamese (ELRA-S0322).

In each language about 100 sentences were read from each of the 100 speakers. The read texts were selected from national newspapers available via Internet to provide a large vocabulary (up to 65,000 words). The read articles cover national and international political news as well as economic news. The speech is available in 16bit, 16kHz mono quality, recorded with a close-speaking microphone (Sennheiser 440-6) and same recording equipment for all languages. The transcriptions are internally validated and supplemented by special markers for spontaneous effects like stuttering, false starts, and non-verbal effects like laughing and hesitations. Speaker information like age, gender, occupation, etc. as well as information about the recording setup complement the database. The entire GlobalPhone corpus contains over 450 hours of speech spoken by more than 1900 native adult speakers.

Data is shortened by means of the shorten program written by Tony Robinson, available from Softsound's web page: http://www.softsound.com/ linux distributions, or simulated versions such as cygwin. Alternatively, the data could be delivered unshorten.

The Turkish corpus was produced using the Zaman newspaper. It contains recordings of 100 speakers (28 males, 72 females) recorded in Istanbul, Turkey. The following age distribution has been obtained: 30 speakers are below 19, 30 speakers are between 20 and 29, 23 speakers are between 30 and 39, 14 speakers are between 40 and 49, and 3 speakers are over 50.

Desktop/Microphone
The Spanish Speech Corpus 1 contains the recordings of 200 native Spanish speakers (100 males, 100 females) recorded in an office and a closed public place, over 4 channels, in a range of low to medium background noise environments (Plantronics Audio 10 (computer/desk mic), Shure SM58 (desk mounted dynamic mic), Shure Beta 53 (headset mic) and Andrea DA-400 (array mic)). The data collection and transcription were performed by Appen (Australia).
Speech samples are stored as sequences of 16-bit 22.05 kHz PCM in uncompressed WAV files.
Each speaker read the following items (prompted):
- 100 command words
- 100 phonetically rich sentences
The following age distribution has been obtained: 75 speakers are between 18 and 19, 114 are between 20 and 30, and 11 are between 31 and 45.
Information about the speakers? place of birth is included.
The database is provided with orthographic transcriptions in SAMPA, including canonical and alternative pronunciation, and syllable, stress and acoustic events markings. All transcriptions were segmented at the utterance (sentence/command word) level, annotated at the word level and checked manually. A pronunciation lexicon including 3,748 headwords (plus variants) is also available.
This database is aimed to be used within speech recognition and voice control applications.

The number of people involving recording: The product totally uses 10 speakers (3 males, 7 females). The speakers have different accent, age, and education background.The recording?fs content: The content includes 4 parts: stock, country?fs name, people ?fnamer30 and the Chinese city?fs name. 30 sentences of stock+10 sentences of country?fs name+30 sentences of people?fs name+10 sentences of the Chinese city?fs name.The capacity of product: The 4 channels are 58 MB, totally 39 hours.The single channels are 147 MB, totally 0.79 hours.
http://www.chineseldc.org/EN/doc/CLDC-SPC-2006-011/intro.htm

The number of people involving recording: The product totally uses 120 speakers (59 males, 61 females). The speakers have different accent, age, and education background.The recording?fs content: 50 speakers: 120 messages for one speaker; 70 speakers: 150 messages for one speaker.The capability of product: The total product data is 945 MB, totally 62 hours.
http://www.chineseldc.org/EN/doc/CLDC-SPC-2006-013/intro.htm

The number of people involving recording: The product totally uses 200 speakers (87 males, 113 females). The speakers have different accent, age, and education background.The recording?fs content: 30 sentences for one speakerThe capacity of product: The 4 channels are 698 MB, totally 46 hours.The single channels are 1746 MB, totally 11.5 hours.
http://www.chineseldc.org/EN/doc/CLDC-SPC-2006-010/intro.htm

The number of people involving recording: The product totally uses 70 speakers (38 males, 32 females). The speakers have different accent, age, and education background.The recording?fs content: Every speaker read 60 sentences of stock..The capability of product: The total product data is 776 MB, totally 5.1 hours.
http://www.chineseldc.org/EN/doc/CLDC-SPC-2006-015/intro.htm

The number of people involving recording: The product totally uses 265 speakers (134 males, 131 females). The speakers have different accent, age, and education background.The recording?fs content: 201 speakers: 30 sentences of stock (2 names for one sentence);64 speakers: 15 sentences of stock (2 names for one sentence).The capacity of product: The total data amount of product is 387 MB, totally 7 hours.
http://www.chineseldc.org/EN/doc/CLDC-SPC-2006-005/intro.htm

The number of people involving recording: The product totally uses 64 speakers (52 males, 12 females). The speakers have different accent, age, and education background.The recording?fs content: 201 speakers: 50 messages for one speakerThe capacity of product: The total data amount of product is 161 MB, totally 3 hours.
http://www.chineseldc.org/EN/doc/CLDC-SPC-2006-007/intro.htm