Older adult Mandarin Chinese data

1. Product Overview

The Chinese speech database for elderly speakers comprises a total of 793 hours of recordings. The recorded language is Chinese, and the recordings were made in China. The transcribed texts are categorized into 12 types, including smart control commands and points of interest (POI). A total of 503 participants aged 55 years or older took part in the recording process. Recordings were conducted in a quiet indoor environment using three different devices: a high-fidelity microphone (44.1 kHz, 16-bit, 266H); an Android smartphone (16 kHz, 16-bit, 264H); and an iOS smartphone (16 kHz, 16-bit, 263H). This database has been professionally transcribed and annotated by skilled linguists and underwent rigorous quality assurance, achieving a text accuracy rate exceeding 97%.

2. Product Catalog Structure

Catalog Structure

Data Directory Structure

└─DOC (Text Description Files)
││├─wav_list.txt (Audio File List)
││├─content.txt (Speech Text Content)
││├─spk_info.xlsx (Speaker Information)
│
└─SPEECHDATA (Data Folder)
│├─0001 (Speaker Folder)
││ ││ 0001_Android_0001.wav (Audio File)
││ ││ 0001.txt (Speech Text Content)

Naming Conventions

CORPUS/USAGE/SPEAKER_NUM/SPEECH_IDe.g./SPEECHDATA/0001/0001_Android_0001.wav

Directory Name · Content · Notes

CORPUS · · Speech Database ID

USAGE · SPEECHDATA · Folder Name

SPEAKER_NUM · 0001 · Speaker Folder Name

SPEECH_ID · 0001_Android_0001.wav · WAV File

3. Text Design

Corpus Construction

The corpus was selected from 12 distinct domains and processed according to established guidelines.

Serial Number · Domain

· Household Appliances Names and Control Commands

· Geographic Information

· Music and Radio Controls

· Digital Device Controls

· TV and Movie Titles

· Finance

· Technology

· Sports

· Entertainment

· Current Affairs News

· English Spelling Practice

· General Texts

• De-identification processing: Removed politically sensitive content, personal privacy information, and material involving pornography or violence.
• Removed symbols such as <, >, [ , ] , ～ , / , \ , =.
• Excluded any content containing languages other than Chinese and English.
• Eliminated sentences longer than 40 characters.
• Standardized formatting.

Text Structure

To ensure balanced phonetic coverage and phoneme distribution, each audio file in this database contains approximately 500 sentences drawn from the corpus, structured as follows:

· Serial Number · · Domain · Number of Sentences

· · · Household Appliances Names and Control Commands ·

· · · Geographic Information ·

· · · Music and Radio Controls ·

· · · Digital Device Controls ·

· · · TV and Movie Titles ·

· · · Finance ·

· · · Technology ·

· · · Sports ·

· · · Entertainment ·

· · · Current Affairs News ·

· · · English Spelling Practice ·

· · · General Texts ·

Total · · Items · ·

4. Data Recording Environment

Recording Environment

A quiet room free from obvious external voices or other noise, with no echoes. Speakers read the prepared texts at their normal speaking pace.

Recording Equipment

The recording equipment includes a high-fidelity microphone and recorder, along with smartphones. The data storage format for this database is 44.1 kHz, 16-bit mono for high-fidelity recordings, and 16 kHz, 16-bit mono for smartphone recordings.

Recording Method

Speakers positioned 20 cm away from the high-fidelity microphone, speaking at normal volume and pace while reading the provided texts. Android and iOS smartphones were placed 20 cm away from the microphone respectively.

5. Annotation and Transcription Guidelines

Transcribers convert audio into written text based on what they hear, striving to maintain complete consistency between the spoken words and the written content. Key principles include: 1) Transcriptions must exactly match the spoken audio—no extra words, omissions, or errors. 2) Numbers should be spelled out in Chinese characters, e.g., “一二三,” rather than “123.” Pay special attention to distinguishing “一” from “幺” and “二” from “两.” 3) English words appearing within sentences should be transcribed phonetically, e.g., “thank you.” Letters pronounced individually should be rendered in uppercase followed by spaces, such as “N B A” or “U F O.” 4) Symbols mentioned in the text should be transcribed according to how the speaker actually pronounces them—for example, “三W点百度点com.” Unpronounced symbols should be omitted. Brand names and proper nouns should follow standard usage, such as “QQ空间,” “iPhone,” or “喜马拉雅.” 5) The completeness of annotations must align with actual pronunciation; no deletions are permitted.

6. Speaker Information

Speaker information records details such as task number, gender, accent region, and age.

Task Number · Gender · Accent Region · Age Range

0001 · Male · Northern Region · 58 years old

Task Number: Each speaker receives one unique task number, corresponding to a single set of recorded texts. Each speaker may participate only once. Gender: 239 males and 264 females. Accent Region: Classified according to the speaker’s native regional dialect—Northern, Southern, or Other. Age Range: Refers to the speaker’s age at the time of recording; all speakers in this database are aged 55 or older.

Sample Data Download

Click to download sample data