Documentation Recording

Recording Environment

Detailed information about the recording environment and session protocols used in the TAPS dataset collection.

Recording Environment Setup

Room Configuration

  • • Semi-soundproof room at POSTECH
  • • Forehead rest for consistent positioning
  • • Reflection filter for ambient noise reduction
  • • DC battery powered system

Equipment Placement

  • • Throat mic on supraglottic area
  • • Acoustic mic 30cm from speaker
  • • Nylon filter for pop noise prevention
  • • Fixed head position setup

Microphone Configuration

Throat Microphone Positioning

The throat microphone was carefully positioned on the supraglottic area to capture vocal cord vibrations effectively.

Proper placement is crucial for capturing both vocal cord vibrations and essential speech formants.

Acoustic Microphone Setup

The acoustic microphone was positioned with specific considerations for optimal recording quality.

  • • Distance: 30 cm from speaker's face
  • • Nylon filter installation
  • • Pop noise prevention measures
  • • Ambient noise reduction setup

Recording Session Protocol

Script Selection

Scripts were extracted from the Korean newspaper corpus provided by the National Institute of Korean Language.

  • • Source: 2023 newspaper articles
  • • Sentence length: 40-80 characters
  • • Various topics covered
  • • 100 sentences per speaker

Session Procedure

  1. Speaker positioning with forehead rest
  2. Equipment verification and signal check
  3. Reading sentences displayed on screen
  4. Individual recording of each sentence
  5. Quality monitoring during recording

Quality Control Measures

Several measures were implemented to ensure high-quality recordings:

  • • Real-time monitoring of recording levels
  • • Background noise minimization
  • • Consistent positioning verification
  • • Manual review of recorded content

For the evaluation set, each utterance was carefully reviewed to ensure the recorded speech accurately matched the intended sentence.