TMASC: Transmasculine Attitude and Speech Corpus

2026-06-15 · Source: Computation and Language · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Data Science & Analytics, Natural Language Processing · Depth: Expert, quick

Summary

The Transmasculine Attitudes and Speech Corpus (TMASC) is a newly introduced multimodal dataset featuring contributions from 196 transmasculine individuals. This comprehensive corpus integrates detailed questionnaire responses, specifically exploring aspects of vocal health, with 66 corresponding audio recordings. The audio component captures diverse vocalizations, including cough and throat-clearing samples, a standardized reading passage, and answers to session-specific questions. The accompanying paper meticulously outlines the corpus's development methodology and the precise data collection procedures employed. To illustrate TMASC's practical utility, three distinct case studies are presented, demonstrating its application in integrating perceptual and acoustic data, identifying group-level vocal characteristics, and calibrating acoustic measurements, all designed to support transmasculine individuals.

Key takeaway

For research scientists developing speech technology or studying vocal health in specific populations, TMASC offers a unique, specialized dataset. You should consider integrating this multimodal corpus to enhance the accuracy and inclusivity of your models, particularly when analyzing transmasculine voices. Utilizing TMASC can improve the calibration of acoustic measurements and help identify distinct group-level vocal characteristics, leading to more precise and supportive applications.

Key insights

TMASC provides a multimodal corpus for research into transmasculine vocal health and characteristics.

Principles

• Multimodal data enhances vocal health research.
• Crowd-sourcing can build specialized corpora.

Method

The corpus was developed by collecting questionnaire responses and 66 audio recordings (cough, throat-clearing, reading, questions) from 196 transmasculine individuals.

In practice

• Integrate perceptual and acoustic data.
• Identify group-level vocal characteristics.
• Calibrate acoustic measurements.

Topics

• Transmasculine Voices
• Speech Corpus
• Vocal Health
• Multimodal Data
• Acoustic Analysis
• Data Collection

Best for: AI Scientist, Research Scientist, Data Scientist

Related on AIssential

• Speech-Guided Multimodal Learning for Vocal Tract Segmentation in Real-Time MRI — Multimodal supervision during training can enhance single-modality inference for challenging medical image segmentation tasks.
• A Multilingual Voice Analytics Module for Contact-Center Hiring — SR-Voice enhances contact-center hiring by analyzing vocal performance across emotion, communication, and rhythm dimensions.
• Phonetic Error Analysis of Raw Waveform Acoustic Models — Raw waveform acoustic models achieve state-of-the-art TIMIT PER, with error patterns revealing specific phonetic class improvements from BLSTMs and transfer learning.
• Dolph2Vec: Self-Supervised Representations of Dolphin Vocalizations — Dolph2Vec, a species-specific SSL model, effectively uncovers fine-grained dolphin communication patterns using a novel, longitudinal dataset.
• How Mistral’s Voxtral TTS is Redefining Multilingual Voice Cloning with a Hybrid Autoregressive and Flow-Matching Architecture — Voxtral TTS uses a hybrid autoregressive and flow-matching architecture for high-fidelity multilingual voice cloning.
• Recurrence-Based Nonlinear Vocal Dynamics as Digital Biomarkers for Depression Detection from Conversational Speech — Depression alters vocal recurrence structure, detectable via nonlinear state-space analysis of conversational speech.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Computation and Language.