CITATION — REFERENCE ENTRY

On Ideal Binary Mask As the Computational Goal of Auditory Scene Analysis — Speech Separation by Humans and Machines

Revision ccade395-105c-4833-b6f8-a44b64dc1a9b · 11/21/2026, 5:18:22 AM UTC
Key
wang2005-ideal-binary-mask
Authors
Wang, DeLiang
Issued
2005
Type
chapter
Container
Speech Separation by Humans and Machines
Publisher
Kluwer Academic Publishers
Publisher place
Norwell, MA
Pages
181-197
Raw CSL JSON
{
  "DOI": "10.1007/0-387-22794-6_12",
  "URL": "https://pnlwang.github.io/papers/Wang05.pdf",
  "page": "181-197",
  "type": "chapter",
  "title": "On Ideal Binary Mask As the Computational Goal of Auditory Scene Analysis",
  "author": [
    {
      "given": "DeLiang",
      "family": "Wang"
    }
  ],
  "editor": [
    {
      "given": "Pierre",
      "family": "Divenyi"
    }
  ],
  "issued": {
    "date-parts": [
      [
        2005
      ]
    ]
  },
  "publisher": "Kluwer Academic Publishers",
  "container-title": "Speech Separation by Humans and Machines",
  "publisher-place": "Norwell, MA"
}

Claims

  1. Major ASA cues include proximity in frequency and time, harmonicity, smooth transition, onset synchrony, common location, common amplitude and frequency modulation, and prior knowledge.
    "Major ASA cues include proximity in frequency and time, harmonicity, smooth transition, onset synchrony, common location, common amplitude and frequency modulation, and prior knowledge."
    Locator: page: 182 (§1 Introduction) · Quote language: en
  2. The ideal binary mask is a binary time-frequency matrix with 1 where the target energy is stronger than the interference energy in a unit and 0 otherwise.
    "an ideal mask is a binary matrix, where 1 indicates that the target energy is stronger than the interference energy within the corresponding T-F unit and 0 indicates otherwise. This definition implies a 0-dB SNR criterion for mask generation"
    Locator: page: 187 (§4) · Quote language: en
  3. Wang proposes the ideal binary mask as a main computational goal of CASA, consistent with characteristics of human auditory scene analysis.
    "My analysis results in the proposal of the ideal binary mask as a main goal of CASA. This goal is consistent with characteristics of human auditory scene analysis."
    Locator: page: 195 (§6) · Quote language: en
  4. The same mixture yields different ideal masks depending on which sound is the target.
    "With the same mixture, the definition leads to different masks depending on what the target is."
    Locator: page: 189 (§4, "Flexibility") · Quote language: en
  5. Auditory scene analysis proceeds in two stages: decomposing the mixture into segments, then grouping segments into streams.
    "The first stage decomposes the acoustic mixture into a collection of sensory elements or segments, and the second stage selectively groups segments into streams."
    Locator: page: 182 (§1 Introduction) · Quote language: en
Available in