[{"@type":"PropertyValue","name":"Format","value":"48,000Hz, 24bit, uncompressed wav, mono channel"},{"@type":"PropertyValue","name":"Recording environment","value":"professional recording studio"},{"@type":"PropertyValue","name":"Recording content","value":"recording corpus reflects the text content of each character"},{"@type":"PropertyValue","name":"Speaker","value":"4 professional CharacterVoice recorded 7 styles, namely criminal subordinate, rough man, little girl, kindly grandma, businessman, grandfather, Non-Commissioned Officer style. Among them, criminal subordinate and rough man are recorded for the same voice actor; The little girl, the grandmother, and the businessman are recorded for the same voice actor. 2 hours/style."},{"@type":"PropertyValue","name":"Annotation","value":"word and pinyin transcription, prosodic boundary annotation"},{"@type":"PropertyValue","name":"Device","value":"microphone"},{"@type":"PropertyValue","name":"Language","value":"Taiwanese mandarin"},{"@type":"PropertyValue","name":"Application scenarios","value":"speech synthesis"}]
{"id":1533,"datatype":"1","titleimg":"https://www.nexdata.ai/shujutang/static/image/index/datatang_yuyin_default.webp","type1":"165","type1str":null,"type2":"219","type2str":null,"dataname":"14 Hours Taiwan Mandarin TTS Dataset – Multi-Style Voices","datazy":[{"title":"Format","desc":"Format","content":"48,000Hz, 24bit, uncompressed wav, mono channel"},{"desc":"Recording environment","content":"professional recording studio","title":"Recording environment"},{"desc":"Recording content","content":"recording corpus reflects the text content of each character","title":"Recording content"},{"desc":"Speaker","content":"4 professional CharacterVoice recorded 7 styles, namely criminal subordinate, rough man, little girl, kindly grandma, businessman, grandfather, Non-Commissioned Officer style. Among them, criminal subordinate and rough man are recorded for the same voice actor; The little girl, the grandmother, and the businessman are recorded for the same voice actor. 2 hours/style.","title":"Speaker"},{"desc":"Annotation","content":"word and pinyin transcription, prosodic boundary annotation","title":"Annotation"},{"desc":"Device","content":"microphone","title":"Device"},{"desc":"Language","content":"Taiwanese mandarin","title":"Language"},{"desc":"Application scenarios","content":"speech synthesis","title":"Application scenarios"}],"datatag":"TTS,Taiwan Mandarin,Multi-style","technologydoc":null,"downurl":null,"datainfo":null,"standard":null,"dataylurl":null,"flag":null,"publishtime":null,"createby":null,"createtime":null,"ext1":null,"samplestoreloc":null,"hosturl":null,"datasize":null,"industryPlan":null,"keyInformation":"","samplePresentation":[{"name":"000488.wav","url":"https://storage-product.datatang.com/damp/product/samplePresentation_ipad/20260922180623/000488.wav?Expires=4102415999&OSSAccessKeyId=LTAI5tEBeSWUJiqjXvBMsxEu&Signature=9ouwHT%2Bk3wH7vVeT6BVMhSuYWWk%3D","intro":"這次#1鑑測#3有人#1受傷#3，按照#1規定#3，要#1追究#1你的#1責任#4。 \tzhe4 ci4 jian4 ce4 you3 ren2 shou4 shang1 an4 zhao4 gui1 ding4 yao4 zhui1 jiu4 ni3 de5 ze2 ren4","size":676500,"progress":100,"type":"mp3"},{"name":"000225.wav","url":"https://storage-product.datatang.com/damp/product/samplePresentation_ipad/20260922180623/000225.wav?Expires=4102415999&OSSAccessKeyId=LTAI5tEBeSWUJiqjXvBMsxEu&Signature=n8WzbsRpJ5dEfNTq%2BRG4vrqmPzs%3D","intro":"容我#1再#1強調#1一遍#3，「簡食#1」是#1集團#1今年#1最重要的#1一個#1專案#1計劃#4。 \trong2 wo3 zai4 qiang2 diao4 yi2 bian4 jian3 shi2 shi4 ji2 tuan2 jin1 nian2 zui4 zhong4 yao4 de5 yi2 ge4 zhuan1 an4 ji4 hua4","size":778062,"progress":100,"type":"mp3"},{"name":"000228.wav","url":"https://storage-product.datatang.com/damp/product/samplePresentation_ipad/20260922180623/000228.wav?Expires=4102415999&OSSAccessKeyId=LTAI5tEBeSWUJiqjXvBMsxEu&Signature=o8rkg2SOuNwkpg%2FIy%2BKW94qSHRI%3D","intro":"我的#1孫子#2最近#1很忙#3，每天#1早出晚歸#1不說#3，連飯#1都#1忘記吃#4。 \two3 de5 sun1 zi5 zui4 jin4 hen3 mang2 mei3 tian1 zao3 chu1 wan3 gui1 bu4 shuo1 lian2 fan4 dou1 wang4 ji2 chi1","size":1315866,"progress":100,"type":"mp3"},{"name":"000003.wav","url":"https://storage-product.datatang.com/damp/product/samplePresentation_ipad/20260922180623/000003.wav?Expires=4102415999&OSSAccessKeyId=LTAI5tEBeSWUJiqjXvBMsxEu&Signature=JCn0XeE4sx8MixM%2BQpP3ZxTmIts%3D","intro":"在#1道上#2，我#1聲名#1遠揚#3，無人#1不知#3，都要#1敬我#1三分#4。 \tzai4 dao4 shang4 wo3 sheng1 ming2 yuan3 yang2 wu2 ren2 bu4 zhi1 dou1 yao4 jing4 wo3 san1 fen1","size":768608,"progress":100,"type":"mp3"},{"name":"000118.wav","url":"https://storage-product.datatang.com/damp/product/samplePresentation_ipad/20260922180623/000118.wav?Expires=4102415999&OSSAccessKeyId=LTAI5tEBeSWUJiqjXvBMsxEu&Signature=WFmczB2mU93C3OfYnzsDmNG2kJ8%3D","intro":"你#1喜歡#1粉紅色的#1書包#3，還是#1藍色的#1書包啊#4？ \tni6 xi3 huan1 fen3 hong2 se4 de5 shu1 bao1 hai2 shi4 lan2 se4 de5 shu1 bao1 a5","size":623900,"progress":100,"type":"mp3"}],"officialSummary":"This dataset contains 14 hours of Taiwan Mandarin recordings from 4 professional voice actors with 7 speaking styles. The styles are criminal subordinate, rough man, little girl, kind grandma, businessman, grandfather and non-commissioned officer. Professional phonetician participates in the annotation. It is ideal for text-to-speech (TTS), expressive voice generation, virtual avatars, and AI speech synthesis applications.","dataexampl":null,"datakeyword":["Taiwan Mandarin speech dataset","Taiwan Mandarin voice dataset","Taiwan Mandarin speech corpus for AI","Mandarin accent dataset Taiwan","Mandarin TTS dataset"],"isDelete":null,"ids":null,"idsList":null,"datasetCode":null,"productStatus":null,"tagTypeEn":"","tagTypeZh":null,"website":null,"samplePresentationList":null,"datazyList":null,"keyInformationList":null,"dataexamplList":null,"bgimg":null,"datazyScriptList":null,"datakeywordListString":null,"sourceShowPage":"speechSyn","dataShowType":"[{\"code\":\"0\",\"language\":\"ZH\"},{\"code\":\"1\",\"language\":\"ZH\"},{\"code\":\"2\",\"language\":\"EN\"},{\"code\":\"3\",\"language\":\"EN\"}]","productNameEn":"14 Hours - Taiwan Mandarin Seven Style Average Tone Speech Synthesis Corpus","BGimg":"brightSpot_audio","voiceBg":["/shujutang/static/image/comm/audio_bg.webp","/shujutang/static/image/comm/audio_bg2.webp","/shujutang/static/image/comm/audio_bg3.webp","/shujutang/static/image/comm/audio_bg4.webp","/shujutang/static/image/comm/audio_bg5.webp"]}
This dataset contains 14 hours of Taiwan Mandarin recordings from 4 professional voice actors with 7 speaking styles. The styles are criminal subordinate, rough man, little girl, kind grandma, businessman, grandfather and non-commissioned officer. Professional phonetician participates in the annotation. It is ideal for text-to-speech (TTS), expressive voice generation, virtual avatars, and AI speech synthesis applications.
This is a paid dataset licensed for commercial use. Ready-made datasets are available for immediate integration into AI projects.
Specifications
Format
48,000Hz, 24bit, uncompressed wav, mono channel
Recording environment
professional recording studio
Recording content
recording corpus reflects the text content of each character
Speaker
4 professional CharacterVoice recorded 7 styles, namely criminal subordinate, rough man, little girl, kindly grandma, businessman, grandfather, Non-Commissioned Officer style. Among them, criminal subordinate and rough man are recorded for the same voice actor; The little girl, the grandmother, and the businessman are recorded for the same voice actor. 2 hours/style.
Annotation
word and pinyin transcription, prosodic boundary annotation
What languages and voice characteristics are covered by Nexdata’s speech synthesis datasets?
Nexdata offers speech synthesis datasets covering a broad range of languages, dialects, accents, and voice types, supported by extensive global language resources. Our datasets include diverse speakers, speaking styles, emotions, and recording scenarios to support natural and expressive Text-to-Speech (TTS) model development.
Can Nexdata customize speech synthesis datasets for specific languages or requirements?
Yes. If our off-the-shelf TTS datasets do not fully meet your requirements, Nexdata provides flexible custom data collection, transcription, annotation, and quality control services. We can customize datasets based on target languages or dialects, speaker profiles, voice characteristics, emotions, speaking styles, recording environments, and data volume.
How does Nexdata ensure the quality and scalability of speech synthesis datasets?
Nexdata applies multi-stage quality control throughout voice data collection, transcription, annotation, and validation. Combined with our extensive language resources and scalable collection capabilities, we can support both large-scale multilingual TTS projects and specialized datasets for specific voices, accents, emotions, or speech scenarios.
Specifications
Sample
Recommended Dataset
