ASR

This dataset was recorded in a sound studio environment, with the participation of 300 speakers, comprising 149 males and 151 females. All speakers involved in the recordings were carefully selected by professionals to ensure standardized pronunciation and clear enunciation. The recorded texts cover information such as key phrases and other relevant data.

This dataset was recorded in quiet office/home environments, with the participation of 150 speakers, including 76 males and 74 females. All speakers involved in the recording were professionally selected to ensure standardized pronunciation and clear enunciation. The recorded texts cover commonly used keywords in voice applications.

This dataset was recorded in a quiet office/home environment, involving 250 speakers, comprising 114 males and 136 females. All participants in the recording were carefully selected through a professional screening process to ensure standardized pronunciation and clear enunciation. The texts for the recordings include everyday language and other relevant information.

This dataset was recorded in a quiet office environment, with a total of 199 speakers participating, including 101 males and 98 females. All speakers involved in the recording were professionally selected to ensure standard pronunciation and clear articulation. The recorded text covers daily conversations and fairy tales.

This dataset was recorded in a quiet office environment, with a total of 575 speakers participating, including 283 males and 292 females. All speakers involved in the recording were professionally selected to ensure standard pronunciation and clear articulation. The recorded text covers wake-up words and other information.

This dataset was recorded in quiet office and home environments, involving a total of 1,232 speakers, comprising 636 males and 596 females. All participants who were involved in the recording process were professionally screened to ensure standard pronunciation and clear articulation. The recorded texts cover a range of information including news and daily conversations.

This dataset was recorded in a quiet office/home environment, with the participation of 102 speakers, including 52 males and 50 females. All speakers involved in the recording were professionally screened to ensure standard pronunciation and clear articulation. The recorded texts cover information such as space navigation command words and phrases.

This dataset is a specialized collection of bilingual Chinese-English speech recordings, tailored to cater to the needs of speech recognition technology development. It is characterized by its unique blend of languages, high-fidelity audio quality, and the diversity of its contributors.

【Product Features】High sampling rate, in-vehicle corpus, indoor quiet collection, multiple scenarios (vehicle control, music, general, map, casual chat scenarios) Can be applied to in-vehicle and other common speech recognition scenarios. 【Audio Parameters】 16k: 1 person 0.5 hours 44.1k: 148 people 74.9 hours 48k：2463 people,1354.8 hours