中英文唤醒词语音数据库

一、Overview

database contains 1,561.12 hours speech data, including 3,936,003 wake-upwords speech files.Database language: Chinese and English ◎Recording area: China ◎Wake-up words for recording: “Hi mia” and the Chinese of “你好，米雅” ◎Speakers: 254 participants ◎Environment: Real home environment ◎Device setup: 7 different positions are set for recording, including: ◎1) Six 16-channel circular microphone arrays (16kHz,16bit) for the far-field recording;2) One Hi-Fi microphone for the close-talk recording (44.1kHz,16bit). database was transcribed by the professional speech annotators with high QAprocess, and the accuracy rate of word is 100%, which could be used in research of voiceprintrecognition, wake-up words recognition and so on.

二、Database structure

File structure

File FolderFolder

│ .pdf（Database specification）
│ spk_info.xlsx（Speaker information）
└─SPEECHDATA（Data folder）
│├─speech（Wake-up words speech）

Rule of the file path

\SPEECHDATA\speech\wav\<speakerID>\<wavID>.wave.g. \SPEECHDATA\speech\wav\145\145_1_recorded0_001_normal.wav

Field · Value · Field meaning

<speakerID> · 145-398 · Speaker ID

<wavID>.wav · <speakerID>_<pointID>_<micID>_<textID>_<speed>.wav · audio ID

\SPEECHDATA\speech\text\<speakerID>.txte.g. \SPEECHDATA\speech\text\145.txt

Field · Value · Field meaning

<speakerID> · 145-398 · Speaker ID

<speakerID>_<pointID>_<micID>_<textID>_<speed>.wave.g. 145_1_recorded2_001_normal.wav

Field · Value · Field meaning

<pointID> · 1-6：circular microphone arrays ID；7：Hi-Fi Mic ID · Point position ID

<micID> · recorded2-recorded17 · Microphone ID of an array

<textID> · 001-160 · Text ID

<speed> · Slow, normal, fast · Speaking speed

三、Text script design

database contains two kinds of wake-up words：

Wake-up word 1 · 你好，米雅（Chinese of “ni hao, mi ya”）

Wake-up word 2 · hi, mia

The distribution of the text script design for database is following:TextID: 001 to 120 is in noisy environmentTextID: 121 to 160 is in clear environmentThe noise information of each speaker is included in the speaker information form (spk_info):

TextID · Environment · Wake-up word and speed

001 to 020 · Wake-up word 1 normal speed ·

021 to 040 · Wake-up word 2 normal speed ·

041 to 060 · Wake-up word 1 fast speed ·

Noisy environment · ·

061 to 080 · Wake-up word 2 fast speed ·

081 to 100 · Wake-up word 1 slow speed ·

101 to 120 · Wake-up word 2 slow speed ·

121 to 140 · Wake-up word 1 normal speed ·

Clear environment · ·

141 to 160 · Wake-up word 2 normal speed ·

四、Recording design

Recording environment

1. The recording environment is real home environment, including the electrical appliance, tableand chair, and so on.2. The speaker stands in a fixed position for recording.

Devices

The recording devices includes Hi-Fi microphone and PDM microphone array.

Recording method

The recordings of each speaker could be cataloged into three subset according to the speakingspeed (normal speed, fast speed and slow speed). We simulated real smart home scenes by addingnoise sources such as TV, music, and background noises to the room. The room setting is shown inFigure 1. The high-fidelity microphone is 25 cm away from the speaker. The circular microphonearrays are placed around the person with a distance including 1m, 3m and 5m from the person.The noise source is randomly placed close to one of the microphone arrays for each speaker. Theroom setting and position information is shown below:

Mic Position ID · Distance (m) · Height(m) · Device

1 · 1.5 · 1 · Microphone array

2 · 1 · 1 · Microphone array

3 · 3 · 1 · Microphone array

4 · 3 · 1 · Microphone array

5 · 3 · 1 · Microphone array

6 · 5 · 1 · Microphone array

7 · 0.25 · 1 · Hi-Fi Microphone

五、Transcription Specification

1. The content of the transcription text is 100% the same as the audio content, redundant, lackingor wrong text is not allowed2. If there is English word in the text, it shall be transcribed according to the pronunciation, e.g. “hi,mia”.3. The completeness for the transcription and real pronunciation in the audio should be the same.

六、Information of speakers

The information of each speaker, which is recorded in spk_info file, contains mission ID, gender,age, noise type, see below example:

Mission ID · Gender · Age · Noise type · Noise Position

145 · Men · 25 · Cartoon · Position 2

Mission ID：Each speaker has only 1 mission id, and each mission contains a unique scriptpackage for recording. A speaker can only complete one mission.Gender: Male/FemaleAge: The speaker’s age at recording timeNoise type: Singing, pure music, TV play, cartoon, fan’s noise, sport rebroadcast and varietyshow.Noise position: Randomly placed close to one of the microphone arrays for each speaker.There are 254 speakers in this database, and all of them comes from the northern China(northern Chinese accent).

七、Copyright

All copyright of this document is reserved by (Beijing Shell Shell Technology Co., Ltd.). owns all the right of the amendment, the update, and final explanation for this document.

数据样例下载

点击下载数据样例