AVID: Aalto Vocal Intensity Database with Speech and EGG from 50 Speakers

Aalto University researchers collected speech and electroglottography (EGG) data from 50 speakers in 2021. The data includes recordings at four vocal intensity levels and is labeled for supervised machine learning tasks. The dataset was published in a 2024 article in Speech Communication.

Use Cases

• Train supervised learning models for vocal intensity classification based on the four intensity categories.
• Analyze the relationship between speech signals and EGG signals for voice production studies.
• Develop audio feature extraction methods for intensity-based speech processing.

Strengths

• Includes data from 50 speakers with balanced gender representation (25 males, 25 females).
• Provides recordings at four controlled vocal intensity levels (soft, normal, loud, very loud).
• Features sentence-wise labeling with 19 labels to support supervised machine learning.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• Last update date is unknown; freshness unverified.

Provenance

Source

Aalto University

Collection Method

Recordings were conducted using a constant mouth-to-microphone distance and a calibration tone.

Time Range

2021