healthnexus tab documentation d4d

Grantor	Grant Name	Grant Number
National Institutes of Health	Bridge2AI: Voice as a Biomarker of Health - Building an ethically sourced, bioacoustic database	3OT2OD032720-01S1

📊

Composition

What do the instances represent?

Instances

ID
instances-voice-derived
Name
Instance
Representation
Voice recordings and corresponding derived data per session/participant
Instance Type
Participants, recording sessions, and task-specific recordings (e.g., sustained phonation); adult cohort in v1.0
Data Type
Derived data only in v1.0: spectrograms (513 x N), acoustic, phonetic/prosodic features, and automatic transcriptions; demographic, clinical, and validated questionnaire responses
Counts
12,523
Label
Clinical cohort membership by disease category (voice, neurological, mood/psychiatric, respiratory); detailed labels vary by disease-specific questionnaires and phenotype fields
Sampling Strategies
1. ID
  sampling-1
  Name
  SamplingStrategy
  Is Sample
  Yes; participants recruited from specialty clinics at five North American sites
  Is Random
  No; condition-based recruitment per predefined cohorts
  Source Data
  Patients presenting at participating specialty clinics
  Is Representative
  Not claimed; targeted clinical cohorts
  Why Not Representative
  Targeted enrollment to capture voice-manifesting conditions
  Strategies
  Deterministic cohort-based inclusion with screening per inclusion/exclusion criteria
Missing Information
1. ID
  missing-1
  Name
  MissingInfo
  Missing
  Some participants may have multiple sessions; completeness may vary by session
  Why Missing
  Operational needs; subset required multiple visits to complete collection

Subpopulations

ID
subpop-1
Name
Subpopulation
Identification
- Adult participants (v1.0 release)
- Disease Cohorts
  voice disorders, neurological/neurodegenerative disorders, mood/psychiatric disorders, respiratory disorders
Distribution
- 306 participants; 12,523 recordings; across five North American sites

Distribution Formats

ID
distfmt-1
Name
DistributionFormat
Description
- Credentialed access via Health Data Nexus
- Distributed files in Parquet (spectrograms), TSV (phenotype and features), and JSON (data dictionaries)

Distribution Dates

ID
distdate-1
Name
DistributionDate
Description
- 2024-11-27 (v1.0 release)

🔍

Collection Process

How was the data acquired?

ID *

bridge2ai-voice-v1-0

Name

Bridge2AI-Voice

Title

Bridge2AI-Voice: An ethically-sourced, diverse voice dataset linked to health information (v1.0)

Description

Bridge2AI-Voice is a comprehensive collection of data derived from voice recordings with corresponding clinical information to enable AI research into voice as a biomarker of health. Version 1.0 provides 12,523 recordings from 306 adult participants collected across five North American sites, selected based on conditions that manifest in voice (voice disorders, neurological disorders, mood disorders, respiratory disorders, pediatric cohort planned for future releases). The initial release contains low-risk derived data (e.g., spectrograms and engineered features) and detailed demographic, clinical, and validated questionnaire data; original audio waveforms are omitted in v1.0.

Language

English

Page

https://docs.b2ai-voice.org

Issued

2024-11-27

Keywords

voice
bridge2ai
audio

Created By

Alistair Johnson
Jean-Christophe Bélisle-Pipon
David Dorr
Satrajit Ghosh
Philip Payne
Maria Powell
Anaïs Rameau
Vardit Ravitsky
Alexandros Sigaras
Olivier Elemento
Yael Bensoussan

Status

bibo:published

Was Derived From

Raw voice recordings collected during standardized clinical sessions; v1.0 distributes only derived data (spectrograms, engineered features, and data dictionaries) with original audio omitted.

Addressing Gaps

ID
gap-1
Name
AddressingGap
Response
Provide a large, high-quality, multi-institutional, demographically diverse voice dataset with linked clinical information and standardized collection protocols, addressing prior limitations of small, non-diverse datasets and heterogeneous protocols.

Relationships

ID
rel-1
Name
Relationships
Description
- Each record links participant_id to session_id and task_name; one row per recording for features; one row per participant for phenotype

Is Deidentified

ID

deid-1

Name

Deidentification

Description

HIPAA Safe Harbor identifiers removed (e.g., names, fine-grained dates, contact and device identifiers, biometrics)
State and province removed; country retained
Free-speech transcripts removed
Original audio waveforms omitted from v1.0 distribution

Sensitive Elements

ID
sens-1
Name
SensitiveElement
Description
- Contains health-related demographic, clinical, and validated questionnaire data (de-identified)

Acquisition Methods

ID
acq-1
Name
InstanceAcquisition
Description
- Standardized clinic-based protocol with demographic and clinical questionnaires and task-based voice recordings
- Tasks include sustained phonation and other voice/speech tasks
Was Directly Observed
Yes (voice recordings, tasks performed in clinic)
Was Reported By Subjects
Yes (validated questionnaires and targeted confounder questions)
Was Inferred Derived
Yes (spectrograms, engineered acoustic/phonetic/prosodic features, and ASR transcriptions)
Was Validated Verified
Validated questionnaires used; standardized multi-site protocol

Collection Mechanisms

ID
mech-1
Name
CollectionMechanism
Description
- Custom tablet application with headset for data collection when possible
- REDCap used for data capture; export and conversion using open-source library (b2aiprep)

Data Collectors

ID
collectors-1
Name
DataCollector
Description
- Project investigators at participating specialty clinics across five North American sites

Ethical Reviews

ID
irb-1
Name
EthicalReview
Description
- Data collection and sharing approved by University of South Florida Institutional Review Board
- Submitted for review to the University of Toronto Research Ethics Board

Preprocessing Strategies

ID

prep-1

Name

PreprocessingStrategy

Description

Raw audio converted to mono
Resampled to 16 kHz with Butterworth anti-aliasing filter
Spectrograms via STFT (25 ms window, 10 ms hop, 512-point FFT)
Acoustic features via openSMILE
Phonetic/prosodic features via Parselmouth and Praat
Additional features via torchaudio

Used Software

ID	Name	URL
software-opensmile	openSMILE	https://audeering.github.io/opensmile/
software-parselmouth	Parselmouth	https://github.com/YannickJadoul/Parselmouth
software-praat	Praat	https://www.fon.hum.uva.nl/praat/
software-torchaudio	torchaudio	https://pytorch.org/audio/stable/

Labeling Strategies

ID
label-1
Name
LabelingStrategy
Description
- Automatic speech transcriptions generated using OpenAI Whisper Large
Used Software
- ID
  software-whisper
  Name
  OpenAI Whisper Large
  URL
  https://github.com/openai/whisper

Raw Sources

ID
raw-1
Name
RawData
Description
- REDCap-based data capture (Bridge2AI Voice REDCap v3.20.0; Zenodo DOI 10.5281/zenodo.14148755); raw audio waveforms not distributed in v1.0

Maintainers

ID
maint-1
Name
Maintainer
Description
- Hosted and made discoverable via Health Data Nexus; project documentation at https://docs.b2ai-voice.org

Errata

Is Tabular

partially (mixture of tabular TSV/JSON and array-based Parquet)

Resources

Description	ID	Is Tabular	Keywords	Media Type	Name	Path	Title
Parquet dataset containing time-frequency representations (spectrograms) for each recording with par...	spectrograms-parquet	False	spectrogram, parquet	application/x-parquet	spectrograms.parquet	spectrograms.parquet	Derived Spectrograms
Tab-delimited file with one row per participant including demographics, acoustic confounders, and re...	phenotype-tsv	True	phenotype, demographics, questionnaires	text/tab-separated-values	phenotype.tsv	phenotype.tsv	Phenotype Table
JSON data dictionary describing columns in phenotype.tsv; each key maps to column metadata with a on...	phenotype-json	True	data dictionary, phenotype	application/json	phenotype.json	phenotype.json	Phenotype Data Dictionary
Tab-delimited file with one row per recording containing features derived from openSMILE, Praat, Par...	static-features-tsv	True	features, acoustics, phonetics	text/tab-separated-values	static_features.tsv	static_features.tsv	Engineered Acoustic/Phonetic Features
JSON data dictionary describing feature columns in static_features.tsv; each key maps to column meta...	static-features-json	True	data dictionary, features	application/json	static_features.json	static_features.json	Features Data Dictionary

Subsets

ID
subset-adult-v1-0
Name
Adult cohort (v1.0)
Title
Adult Cohort Subset
Description
Initial public release contains adult participants only; pediatric cohort planned for future releases.
Is Data Split
False
Is Subpopulation
True

Motivation

Composition

Collection Process

Uses

Distribution

Maintenance