TECH TRANSFER · Yindouji
Making sound observable information
Turning real sounds into information that can be recorded, segmented and understood.
Yindouji · WeChat mini program
One recording at a time, into sound research
Yindouji explores the recording, segmentation and presentation of body-sound information: from heart and breath sounds to infant cries — time-labelling sound, extracting acoustic features, and explaining results in plain language.
YindoujiNow live as a WeChat mini program: open WeChat, search "音豆纪" (Yindouji); available modules follow the current mini-program version.
On a phone, press and hold the QR code to save it;
Save the QR code to this device ↓
Concept diagram · not a product interface
Sound recording
Time segmentation
Acoustic information interpretation
01 · WHY SOUND
Why sound is worth studying
Sound is mechanical vibration travelling through a medium: the source displaces particles from equilibrium, elastic coupling passes the disturbance onward, and a longitudinal wave of alternating compression forms. Frequency sets pitch, sound pressure sets loudness, and waveform and spectrum carry the sound's "face".
Different sources leave different acoustic fingerprints: heart sounds come from valve closure with clear time–frequency features; airflow in the airways produces laminar and turbulent regimes; vocal-fold vibration sets a fundamental, shaped by the vocal tract into intelligible speech. These differences are the physical basis of "reading information from sound".
Once recorded, time- and frequency-domain analysis turns a clip into comparable numbers: duration, F0, energy, spectral centroid… Researchers form hypotheses and test methods on them — the starting point of Yindouji's exploration: making sound recording and analysis belong not only to expensively equipped labs.
02 · SOUND MAP
Sound research map
Yindouji's sound content spans body sounds, animal sounds and natural soundscapes; what is listed is research content,not equal to a live detection feature of the mini program。
The eight entries are "sound type × research application" combinations: two are voice applications; breath sounds and cough are counted together.
In depth: from recording to comparable features
After recording, common analysis dimensions include time domain (duration, peaks, envelope), frequency domain (F0, formants, spectral centroid) and stability indicators (jitter, shimmer, harmonic-to-noise ratio). Their meaning differs by source: heart sounds care about S1/S2 timing, breath sounds about inspiration/expiration phases, voice about vibration quality.
Methodologically, the classic pipeline is "capture → feature extraction → model classification/segmentation → comparison against a reference standard". Yindouji-related research has built and validated automatic time-label segmentation for heart sounds, breath sounds and infant cries (see "Research cases").
Animal-sound work falls into three uses: ecological monitoring (birds, wildlife), companion-animal health (cats, dogs) and livestock management (pigs, cattle, sheep, poultry…).
In depth: three levels of animal-sound research
hearing— where the target sound is recorded: pen, field or home; microphone distance, ambient noise and sample rate together decide recording quality.Resolution— isolating a target individual or event from a mixed soundscape: species identification and observation of states such as oestrus, stress or pain.Interpretation— linking acoustic differences to physiological or behavioural states, with scope and limitations stated.
Natural-soundscapes research studies sound and environment: habitat quality, seasonal and circadian rhythms, and noise effects on animal communication. White noise and human relaxation responses are related threads (papers under "Research sources & FAQ").
03 · HOW TO USE
User guide
How to choose and record a task, and what results look like — three steps to get started.
Enter
Open WeChat and search "音豆纪" (Yindouji) to enter.
Entryrecord sounds
Choose a sound task in the mini program and record as prompted; distance, duration and environment requirements follow the task page.
Experienceview results
View segmentation and results in the mini program; the five result forms are described below.
Experience04 · RESULTS
Reading the results: five forms at a glance
Yindouji presents results in five forms; what to look for in each:
| Result types | What to look for |
|---|---|
| TXT report | Record fields, label meanings and units; read alongside the task description. |
| Static images | The axes of waveform/spectrum (time, frequency) and the annotated sound segments. |
| Animated images | Which element changes over time and how it maps to the original sound. |
| Excel data | What each row means (a cycle/event) and each column's field and unit. |
| PDF report | Report structure, result sections and interpretation; results are information, not diagnoses. |
To see it for real, the quickest way is to run a task in the mini program.
05 · ONLINE TOOLS
Online apps & intelligent analysis
Around "recording and researching sound", the online tools aim to let part of the capture–organize–present chain happen right in the browser.
Sound task management
Organize recordings, annotations and curation by sound type, supporting longitudinal research material.
Directionsegmentation & feature display
Automatic segmentation and acoustic features on a timeline, assisting manual review.
validated in researchResearch assistance
Research-assistance tools for design, randomization and data organization, complementing the main Yindouji system.
Related tools06 · RESEARCH CASES
Research cases
The following three studies, published in peer-reviewed journals, are validated work behind Yindouji's "automatic sound time-label segmentation" direction. Content is based on paper abstracts and locally held full texts;The findings are academic results — not implemented Yindouji features, and certainly not for individual diagnosis。
Automatic time labels for heart sounds
The study built and validated automatic time-label segmentation of heart sounds: splitting the cardiac cycle, identifying S1 and S2, with pulse-oximeter heart rate as the validation reference.
Evidence: locally held full text (abstract and dates verified) · a methods validation study, not a diagnostic tool.
Inspiration/expiration segmentation of breath sounds
Breath sounds from 84 patients at six chest landmarks were collected with an electronic stethoscope; an automatic inspiration/expiration segmentation method was built, validated and compared across conditions.
Evidence: locally held full text · the paper acknowledges Chengdu Zhiju Data Technology Co., Ltd. for professional discussion.
Automatic time-label segmentation of infant cries
Across 226 infant-cry segments, an automatic inspiration/expiration segmentation method was built; duration, frequency and intensity features were extracted and compared between preterm and full-term backgrounds. Note: 226 counts segments.
evidence: local full-text papers · The paper acknowledges Chengdu Zhiju Data Technology Co., Ltd. for professional discussion. As it involves infants, no screening/diagnostic claims are made.
RELATED TOPICS
Understanding research through body sounds
The entries below open research and visual-creation features for a quick overview.
07 · SOURCES & FAQ
Research sources & FAQ
Research directly related to the Yindouji direction
- Construction and validation of an automatic heart-sound time-label segmentation method · Frontiers in Artificial Intelligence (2024) · doi.org/10.3389/frai.2023.1309750
- Automatic time-label segmentation of breath sounds · The Clinical Respiratory Journal (2025) · doi.org/10.1111/crj.70124
- Automatic time-label segmentation of infant cries · BMC Pediatrics (2025) · doi.org/10.1186/s12887-025-06071-z
- Sedative effect of white noise on the prefrontal cortex (fNIRS RCT) · Noise & Health (2025) · doi.org/10.4103/nah.nah_88_24
- Deep magnetic stimulation and the sleep-deprived brain–heart axis · Frontiers in Neuroscience (2025) · doi.org/10.3389/fnins.2024.1464299
FAQ
Can Yindouji diagnose disease?
What equipment do I need?
Any requirements for recording?
How to read the results?
Where is the entry point?
How is my recording handled?
CONTACT
Explore more possibilities of sound with us
For field trials, joint research or tooling, get in touch.
Contact Usor call us directly at 028-6744 0718